What's new

This page tracks meaningful changes to Evergrowth’s product documentation so you can see what is new or different since you last looked. It is kept current automatically as the product evolves.

2026-10

• Very large accounts are capped at 500 contacts per workflow run. When a workflow reaches an account that has 500 or more of its contacts in the same run, it leaves that account alone and marks those contacts as skipped, rather than taking on one enormous account in a single pass. Selections spread across many companies are unaffected; for one very large company, launch its contacts in batches of under 500. See An agent did not run on a record.
• A run’s credit limit is now a hard ceiling. Before each paid step, a run with its own Credit limit sets aside the most that step could cost and only goes ahead if it still fits, so the run never ends above the limit - at most a little below it. When you raise the limit to continue, leave room for the work still to do. A contact search interrupted by the limit only looks for the contacts it still needed when you resume. See Cap what a run can spend.
• See what a run researched and what it reused. In a record’s Per-record breakdown, a research step’s Research results now separate what this run researched (and did not find) from what was already researched by an earlier run, with each answer marked Researched this run or From a previous run. See Monitor running agents in the task feed.
• Qualification only looks for websites when you ask it to. Launching account qualification, from the qualify window or through Eva, now adds the website-finding pass only when Automatically find domain if missing is on. See Run a Qualification agent.
• Eva answers when @mentioned in Microsoft Teams channels and group chats. Wherever she is a member of a team or a group chat, a mention now gets an answer in that thread, under the asker’s own access and credits, and she introduces herself when added: she answers only when mentioned, and reads only threads she has been mentioned in. The package on the Integrations page is still for direct messages; channels for most organizations arrive with the Microsoft Teams Store listing. See Channels and group chats.
• Act on a saved view without finding it first. Once a view is applied, a menu joined to the Saved Views button offers Edit saved view, Share, favourite and Copy link for that view. Searching the dropdown now covers every tab, shows how many views match in each, and jumps to All when the open tab has none. See Act on the view you have applied.
• Paste a full website address as a value proposition source. The Add sources window now accepts whatever your browser’s address bar holds and reduces it to the site’s domain, instead of asking for a bare domain. See Generate from sources.
• The CRM’s own error in the sync log. When your CRM refuses a write, the record’s sync reason now reads “Your CRM returned an error:” followed by the CRM’s message - a missing required field, for example - and a failure on Evergrowth’s side is labelled as a system error, so you can tell which side needs the fix. See My record will not sync to the CRM.
• CRM syncs leave more of your CRM alone. In Salesforce, an update to an existing lead no longer re-runs your lead assignment rules, so the lead keeps its owner. A fixed value set to fill only empty fields no longer overwrites a dropdown value it does not recognise. And Attio fields for plays and account plans are created as plain text, so no HTML tags show up in them. See Connect your CRM.
• Workflows can tag records and fill in custom fields. The workflow builder has a new Actions group with two free steps. Add tags puts the tags you choose on every account or contact that reaches it, and Set attribute writes a fixed value into one of your custom fields. Drop one after the branch that matters - qualified accounts, contacts who got plays - and the run leaves behind a list you can filter, save as a view or export, with no credits spent on the labelling. See Action steps.
• Launch agents on a batch from the extension. The accounts and contacts lists in the browser extension, including the records you have just captured from a page, now offer the same agents as the workspace’s lists for every record you select, next to launching a workflow. See Running agents and the Chrome extension.
• Put a credit cap on a single run. Every launch window - each agent’s and Launch workflow - now has a Credit limit switch under the estimate, and the Workflow Scheduler’s Target & limits step has Never spend more than, a cap in credits per run. A run that reaches its limit pauses as Workflow limit reached and keeps what it finished; Raise limit & continue or Remove limit & continue picks it back up. You can also set or change the limit on a run that is already going, from the credits figure in its run report. See Cap what a run can spend.
• Workflows can refill the seat a contact left. The Contact Qualification Agent step in the workflow builder has a new switch, Find a replacement for contacts who left. When a contact turns out to have left their company, the workflow searches their old account for someone with the same persona (or the same job title, if they had no persona) and carries a replacement who fits on through the rest of the workflow. It works on its own or alongside Follow contacts who changed jobs, so one departure can both follow the person to their new company and refill their old seat. The search shows as a Find replacement step in that contact’s per-record breakdown. See Build a workflow.
• Website Finder costs less, and has a Smart Mode. Website Finder now costs 0.3 credits per account at Low Effort and 0.5 at High Effort. A new Smart Mode runs the quick search first and the deep search only when nothing was found, so it finds about as many websites as High Effort for around 0.37 credits per account on a typical list. It is offered wherever you choose the agent’s effort: the Website finder window, the website option in an account qualification, and the Website Finder step in a workflow. See Find or fix a company website.
• Stop any agent run at a target. Every launch window that works on more than one record - the qualification, research, account plan, play, Website finder, Contact finder and Email/Phone finder agents as well as Launch workflow - now has a Stop early at a target card. Switch it on, say how many qualified accounts, researched contacts, contacts with an email or other result you need, and the run ends once it has them; the records it did not reach stay untouched and cost nothing. The launch button repeats the target, account targets can start from the weekly number on your profile, and a workflow’s own target (the builder’s Target tab, and the scheduler’s Target & limits step) is where every launch of it starts. See Stop a run early at a target.
• Eva in Microsoft 365 Copilot. A Microsoft 365 Copilot card joins the Eva in your AI assistant section on the Integrations page. Your Microsoft 365 admin uploads the Evergrowth package once in the Teams admin center, then everyone opens Evergrowth under Agents in Copilot, on the web or in Teams, and signs in. The free Copilot Chat is enough, and Eva’s answers arrive in the same live panel as in Claude and ChatGPT. See Ask Eva from Claude and other AI assistants.
• The task feed says which credit ceiling stopped a run. A run stopped on a person’s or a team’s credit limit is now labelled User limit reached or Team limit reached instead of sharing the Out of credits label with the workspace balance, and the wording names who it applies to - you, your team by name, or the colleague who launched it. See A run stopped because a credit limit was reached.
• Codex gets its own card, and coding assistants can set Eva up themselves. The Eva in your AI assistant section on the Integrations page now has a Codex card beside Claude Code. The Claude Code, Codex and Other AI tools setup guides open with a prompt you copy and paste into that assistant: it adds Evergrowth for you and ends by telling you to sign in. Every guide now shows the connector link to copy and a drawing of where it goes, Cursor has a one-click Add to Cursor, and if you already added Evergrowth in Claude, Claude Code has it too. See Ask Eva from Claude and other AI assistants.
• People between jobs are flagged in the extension. When you add contacts from LinkedIn through the browser extension, anyone whose profile lists no current company is marked No current company and cannot be selected, because a contact needs an account to belong to. See Create a single account or contact manually.

2026-09

• Eva in ChatGPT, and a one-line setup for Codex. Evergrowth can now be added to ChatGPT as an app, and Eva answers there in the same live panel as in Claude, with follow-ups continuing the same conversation. The ChatGPT card on the Integrations page has the four-step setup; on ChatGPT Business, Enterprise and Edu a ChatGPT admin adds it once and publishes it, then everyone connects. The Other AI tools card now also shows the single command that adds Evergrowth to Codex. If a panel cannot show an answer, it now offers to open that answer in Evergrowth, and a short interruption shows Reconnecting to Evergrowth… instead of giving up. See Ask Eva from Claude and other AI assistants.

• Boost a single run. The launch windows for qualification, research and account plans, and the Launch workflow dialog for workflows that include them, have a Boost switch in the header. It runs that one launch at Fast speed - the qualification and research work in it costs double, on a faster lane - without changing any agent’s saved settings. The estimate reads (2x boosted) so the price is clear before you start. See Boost a single run.

• Ask Eva from Claude, Claude Code, and other AI assistants. Evergrowth can now be added as a connector to the AI assistant you already use, so a question about your accounts, contacts or buyers asked there is answered by Eva from your workspace rather than from the open web. In Claude the answer arrives in an Evergrowth panel laid out as it is in the workspace; in Claude Code and other tools it comes back as text. Each person connects their own assistant and signs in as themselves, and the same access rules and credits apply. The cards are in a new Eva in your AI assistant section on the Integrations page, which is now grouped into sections with tabs; ChatGPT is listed as coming soon. See Ask Eva from Claude and other AI assistants.

• Share a saved view in two clicks. Every view in the Saved Views dropdown now has a Share action that copies a link opening the list with that view applied, or assigns it to colleagues - no need to open the edit dialog to share. See Share a view.

• Only one Contact Finder per account at a time, and a Condition runs every branch that matches. A Condition in a workflow is not first-match: a record goes down every branch whose filters it meets, all at once. Where two of those branches each lead to a Contact Finder, or one Contact Finder leads on into another, the account would be searched twice, so the second step now stops with a plain reason in the run report instead of an unexplained failure: combine the personas into one Contact Finder step, or chain the alternatives so each account reaches only one. See When a record matches more than one branch.

• Zoho leads sync. Leads in Zoho now sync as their own records, the same way they already do for Salesforce and Microsoft Dynamics 365. A lead your team already converted in the CRM is skipped with that reason rather than failing the update. See Integrations.

• A full CRM says it is full. When a field Evergrowth would create in your CRM cannot be added because the CRM has reached its limit for custom fields, you are told exactly that, and what to do about it, instead of a generic “try again later”.

• Refunds and late charges show up in your usage report. Credits handed back and usage billed after the fact now count in the report’s used figure and all-time spend, and draw on your balance oldest credits first, like any other spend. They belong to no agent, so the per-agent shares are now worked out of agent spend alone and always add up to the whole. See Read the credit usage report.

• Your actions follow you down the page. On an account or contact page, once the header scrolls away a slim bar pins to the top with the record’s name and the same Strategize with Eva, Launch agent, Launch workflow, and … actions; click the name to jump back up. See Account detail page and Contact detail page.

• What does Accept-all actually mean? Each check result in the Saved emails card now has an information icon that explains it in one line - an Accept-all address may deliver, but bounces are more likely. See Record status fields.

• A Created at column on the Contacts list. Sort it newest first to review exactly what the last import, sync, or Contact Finder run brought in. See Contacts list: columns and filters.

• The extension panel can close itself, and shows your credits. The extension’s side panel now has its own close button, which matters in browsers such as Edge that draw the panel without one, and your picture in its header wears a ring showing how much of your tightest credit limit is left. Click it for the full credit breakdown, Buy more credits, and Log out. See Find your way around.

• A profile page of your own. Profile settings in the account menu opens a page for your name, what Eva calls you, your role, your timezone, and a photo - and the sidebar now shows your name and photo instead of your email address. Add your LinkedIn and the workspace reads your photo and background from it, then works out your own DISC profile: a suggested tone of voice, your leading traits, and do’s and don’ts, the same read you already get on the buyers you sell to. Sellers can also set their weekly rhythm - new accounts a week, contacts per account, or a named list - so Eva sizes what she suggests to the work you actually take on. Eva addresses you by the name you give her. See Set up your profile.

• Classic workflows are now Legacy, and disappear when you have none. The older builder’s card in the Workflows hub is badged Legacy and its list reads Legacy Workflows. The card only shows while your workspace still has workflows built there, so a team that never used it, or has finished moving across, no longer sees it at all. The retirement date is unchanged: the end of 2026. See Workflows.

• See a workflow before you launch it. Hovering a workflow in the Select workflow dropdown now previews Workflow Builder workflows too, drawing the step chain with its branches, with the description and tags beneath. Picking the wrong one of two similar names on a large selection is now a hover away from being caught. See Run a workflow on records now.

• Templates sorted by where they start. The template gallery now has Account templates and Contact templates tabs in place of the team and Evergrowth groups, and each card shows the icons of the agents it uses. See Start from a template.

• Star workflows from the extension too. The favourite star now works in the extension’s launch dropdown as well as in the workspace, and the same favourites lead the list in both. Scheduling a run still happens in the workspace.

• A workflow on pending contacts points you at the fix. Starting a workflow on contacts that are all still waiting for review is refused with a card that says so, and its Review pending contacts button opens the review already narrowed to those contacts. Approve or place them and launch again. See A workflow will not start on pending contacts.

• A nudge to install the extension. On a desktop browser that can run it, a small notice at the bottom left offers the Evergrowth extension, naming your browser on the button, and confirms once it is installed. It stays away for a week after you close it, then every two weeks, and never shows if you have used the extension in the last two weeks. See Find your way around.

• Persona cards have names. A card now carries a name you choose - Finance Fiona, RevOps Riya - shown as its title with the expertise beneath, searchable, and unique in your workspace. It is what finally lets two cards that look alike on paper be told apart, in the list, in conversation with your team, and when you ask Eva for “the contacts of Finance Fiona”. Eva names every new card she creates; the moment you set or change a name yourself it is yours, and she will not rename that card unless you ask her to rename that one. Renaming changes nothing but the name - assignments, attached research agents and the per-vertical personas all stay as they are. Cards from before this change are simply unnamed and can be named at any time. See Persona card fields and Create persona cards.

• A thumbs-up now promotes an email or number to the main one. The thumbs at the end of each row in the details card move the record, not just the feedback log: thumbs-down still pushes a bad value down, and thumbs-up pins that value as the one the record leads with, the same as Use as main. After a call, the fastest fix is to thumb up the number that worked. A locked main value is left alone either way - unlock it first if you want to replace it. See Find verified email and phone.

• A locked email is always the one shown. An address you locked now stays on the record and counts as a usable email in the Has email filter whatever a verification service says about it, including Invalid - the lock is your decision and it outranks the verdict. Relatedly, a check that comes back in terms the workspace does not recognise now reads Not checked rather than hiding the address as invalid. See Record status fields.

• A No on the IBP chip says why on the chip. Where a reason was recorded, the contact’s IBP chip reads No: Left the company, No: Employment unverifiable, No: Expertise mismatch, No: Seniority mismatch, or No: Technical issue, so a list of rejected contacts can be read down the column without hovering each one - and the three different jobs hiding behind “No” (re-find the person, revisit the persona, requalify) separate at a glance. See Contacts list: columns and filters.

• Control tower stops treating an inactive teammate as a setup gap. Somebody who has not been in the workspace for a fortnight now sorts below the real problems, is left out of the no-seat count, and reads Inactive user, nothing to set up instead of being offered an ownership rule that would route fresh work to somebody who is not there. The exception is the CRM seat, which stays offered with its reason: a rep who works only in their CRM still owns records there. The ownership check also says what a missing rule really costs - new records stay with whoever added them or ran the workflow, and nothing is shared out across the team. See Control tower checks.

• Removing a shared widget from your own dashboard says what it does not cover. Delete only for me drops the share made to you personally; where the same widget also reaches you through a team you belong to, it keeps showing until that team assignment changes. See Use velocity widgets.

• Control tower separates what is broken from what is merely worth knowing. A fifth result, Heads-up, nothing to fix, marks the things no settings screen can close: the play agent or the account planning agent having nothing to work with, or a CRM channel that used to arrive and has gone quiet. Heads-ups take no fix number and are not counted among the gaps, so a workspace that simply does not use plays now finishes its setup instead of carrying a permanent red mark. The play check also relaxed - one ongoing play of any type is enough, where it used to ask for both an email and a call play - a new check watches whether the account planning agent has a plan to generate from, and the separate play-permissions check was dropped, since role permissions belong in team settings. Hovering any dot in the banner’s bar now names its check and its result and clicking one jumps to the row. See Control tower checks.

• Star the workflows you actually run. Every workflow card, in both builders, carries a star that pins that workflow to the top of your list, and the same favourites lead the Select workflow dropdown when you launch one on records. Favourites are personal: starring a workflow reorders your list and nobody else’s, so two people can each keep their own handful at the top of the same shared library. See Keep the ones you use at the top.

• Records pushed in by your own automations now say so. An account or contact created over the API by a webhook or automation used to be labelled Manual, as though somebody had typed it in. It now carries a Webhook source, so the Source column tells you where a record genuinely came from. Note that the Manual / imports quick filter does not catch these, by design. See Contacts list: columns and filters.

• A contact qualified in a classic workflow gets its digital twin too. Qualifying a contact has always been what produces their LinkedIn insights, DISC profile, and persona card, but a contact run through the older classic builder came out a persona fit with all three empty. Classic workflows now do the same work the Workflow Builder does, so the two behave alike and nothing is left half-done. See Digital Twin.

• A refused change to a saved view says who can make it. Editing a view or a view group you do not own used to fail with an error that carried no reason. It now reads “Only the owner of this saved view or an admin can change it”. See Save and recall filter views.

• Control tower: one page that says what your workspace still needs. A new sidebar entry runs a fixed set of checks against your workspace every time you open it - CRM connected and mapped, activity arriving, records owned, ecosystems and personas defined, qualification and research agents in place, credits in order - and puts the failures in the order they have to be fixed, each with a button to the screen that clears it. The first blocker is promoted as Next up, passing checks fold away, and where the fix is thinking work rather than plumbing, a button beside it hands the job to Eva, already pointed at that exact gap. A Team readiness table below shows who has no CRM seat, who owns nothing, and who has gone inactive. Admins always have the page; everyone else sees it when at least one failing check is theirs to clear, and a coloured dot on the sidebar entry says whether anything is wrong before you open it. See Check what your workspace still needs and Control tower checks.

• One verdict per email and phone, and everything else in one card. The reach row no longer spreads a status icon, a source icon, a location mark and four buttons across the slot. Each channel now leads with a single icon whose colour is the answer to “can I use this” - green and ticked when it is checked and good, orange when it is worth a look first, red when it should not be used - and on a verified number that colour is the number’s intent priority. Hover the value and a card lists every address or number the contact holds, each with its check result, where it came from, when it was found and checked, and the location comparison; the actions live there too - verify or re-check, edit, promote another value with Use as main, push a bad one down, and Find more emails or Find more phone numbers at the bottom. Promoting past a locked value now reads Unlock to replace, so nothing overwrites a value your team protected without you saying so. See Contact detail page and Find verified email and phone.

• A channel that came back empty leads with the alternatives. Where a search ran and found nothing, No valid email found or No phone numbers found is itself the trigger: hover it for when the search ran and what to do instead - search again across all providers, add the value you already have by hand, open their LinkedIn, or find someone else at the account. The row no longer re-offers the lookup that just failed. See No email or phone found.

• Retry the records that errored, from the run report. A run with failures carries a banner naming what failed and on which step, with Show failed to isolate those records and Retry errors to re-queue just them - everything that already succeeded is left alone. Because retrying re-runs work and draws credits again, the button is shown only to people whose team can launch workflows. See Retry the records that errored.

• The credit estimate prices the steps a launch implies. Account plans need qualification and research to have run; an account qualification with Automatically find domain if missing switched on needs a Website Finder pass. Those are added for you at launch, and now they are in the estimate as well, so the quoted number is for the run that actually happens rather than for the steps you happened to list. See Seeing the price before you launch.

• One-click quick filters on the Accounts and Contacts lists. A Quick filters bar now sits above both lists, with a chip for each question those lists get asked daily and a live count on every chip - accounts nobody has qualified, contacts with no email, the people who have left the company. Chips answering the same question replace each other, chips answering different ones stack, and Clear all drops the lot. The … menu pins the chips you want, renames them to your team’s words, and saves your current filters as a personal shortcut. See Filter accounts and Filter contacts.

• Show only the records one run touched. The Recent tasks chip opens a picker of your own or the team’s recent agent and workflow runs, searchable by agent or workflow name, and narrows the list to the records that run worked on. Your other filters stay applied on top, so “what that run found, that also qualified” is two clicks. See Filter contacts.

• The filter panel opens on the tab that holds your conditions. Conditions applied while the panel was shut - a quick filter chip, a recalled saved view - used to leave you looking at an empty Basic tab. The panel now opens on Advanced when that is where the applied conditions live, and never switches tab under you while it is open. See Filter accounts.

• The Domain Finder agent is now the Website Finder agent. Same agent, same work, same credits - the name just says what it gives you rather than what it is called in a database. It reads Website finder agent in the agent picker and the launch window, Website Finder Agent as a workflow step, and Website finder in the task feed filters and the credit breakdown. The field it fills in on the record is still Domain. See Website Finder agent.

• An account with no website found says so, and offers to try again. Where the agent has run and come back empty, the Domain cell reads Domain not found in amber instead of quietly showing nothing; hover it for why the search failed and a Find website button to run it again. Accounts it has not tried yet still carry the button outright. See Find or fix a company website.

• A scheduler’s limit is a pace, not a tripwire. Process at most used to pause an automation the moment its audience matched more records than the number allowed, which could leave a job producing nothing for weeks. It is now a per-run work budget: the extra records wait their turn and go out on the following runs, oldest-waiting first, so a wide audience is worked through instead of blocked. Nothing is skipped and nothing pauses. A genuinely enormous audience - far above the 10,000 default - still pauses and asks you to narrow it. If you set a small limit some time ago to be careful, check it: it is now pacing the work rather than guarding against it. See Set up a Workflow Scheduler automation.

• Run history now shows what a run worked on, not what it matched. The Processed column was reporting the size of the audience, so a run that matched 34,000 records and started none of them read as 34,000 processed. It now shows the records the run actually took on, with the matched total underneath when the two differ. The automation’s own panel names the number as Work per run and says when it is the default. See Set up a Workflow Scheduler automation.

• Ask Eva for help in Microsoft Teams. Send help in her chat and she answers straight away with what she can do and how to point her at a workspace. It is a fixed reply rather than a turn she reasons about, so it costs nothing. See Talk to Eva in Microsoft Teams.

• A bulk action clears the rows it ran on. Tag a hundred accounts, launch an agent on them, change their owner - once the action goes through, the selection empties, so the next thing you pick starts from nothing. Cancelling a dialog still leaves your selection where it was. See Bulk-edit records.

• Agents is now Agent Hub. The sidebar entry that opens your agent library was renamed; the lists inside it, and everything you do there, are unchanged. The permission that governs the area is still called Agents in team settings, so the two read differently for a moment. The sidebar icons were redrawn in the same release, agent avatars included. See Find your way around the workspace.

• Eva in Teams is direct messages for now. Channels and group chats are held back until the Microsoft Teams Store listing, which is also what will let her read the rest of a thread - answering in a conversation she can only see half of is the thing worth avoiding. A channel mention gets a short pointer to her direct chat, and nothing is charged for it. If your team’s selling conversation happens in channels, connect Slack, where that already works end to end. See Talk to Eva in Microsoft Teams.

• A phone number now says where it is registered. Beside each number sits a location mark: hover it for the country - the state or metro for a United States number - its dialling code belongs to, compared with the contact’s own location. A number registered somewhere else turns orange, which is a prompt to check before you dial rather than a claim the number is wrong, since people keep their numbers when they move. It shows in the Phone column, on the contact’s Reach row, behind the +n link, and in the browser extension. See Where a number is registered.

• Give Eva a LinkedIn profile link and she adds the contact. No name, no company, no account needed first: she reads the profile for the person’s details and creates or matches their employer from it. Hand her several links at once and she adds them together, telling you which she could not read instead of guessing a name. See Eva.

• A channel you switched off in the finder stays switched off. Some contact-data vendors answer for email and phone in one go whether or not you asked for both. A phone-only or email-only run now discards the channel you did not ask for rather than writing it to the record, so a targeted run cannot replace an address or number your team curated. See the waterfall.

• The task feed card carries what the run cost. Credits moved out of the result chips and into the line of detail under each card, next to who started it and when - so the cost sits with the run rather than competing with its results. The header panel is now titled My last agent runs and leaves the name out, because everything in it is yours. See Monitor the task feed.

• Opening a contact finder run lands on the contacts it found. The account’s Contacts tab now arrives filtered to that run, so a few new people are not lost among the hundreds already on the account. See Opening the record a task ran on.

• A contact’s relationship attribute is labelled Relationship. It read Strength before, which collided with Profile Strength in conversation. The six values and the menu behind it are unchanged. See Contact detail page.

• The credits meter now measures this billing period, not your whole contract. The header chip and the workspace row behind it read your balance against the allowance your plan grants each period, so a healthy workspace stops drifting toward red simply because it has been used for a year. Credits carried over from earlier periods show as their own segment past the end of the allowance, and the bar only turns red at a tenth or less remaining - draining an allowance is the product working, not a warning. See What the credits meter is measuring.

• A filtered list that finds nothing now shows you why. Every table and list in the workspace replaces the bare “no results” with the filters that produced the empty result: removable chips for each one, the overflow behind a +N more filters pill, advanced conditions shown as readable logic, and Clear all filters to start again. On an account’s Contacts tab it goes further and offers the agents that fix the usual cause - Qualify N existing contacts and Find more contacts - counted across the whole account. See Filter accounts.

• A finished run now opens the work, not a report about it. Hover a task-feed card, or read the pop-up when a run lands, and the first action goes straight into the output: one record opens its page on the tab the last agent wrote to, a batch opens the Accounts or Contacts list filtered to that run. Failed runs still lead with See what failed, and a run stopped for credits with Run details. Analyse with Eva now sits on the card beside it rather than inside the report menu. See Opening the record a task ran on.

• The Users list shows who is actually using the workspace. A Last active column joins Created, both reading as an age with the exact date on hover, and the list now sorts by name, created, or last active - opening on last active so the dormant seats fall to the bottom. An empty Last active is somebody who accepted an invitation and never came back. See Manage users.

• Domain is no longer required to import contacts. A contact CSV needs a Contact LinkedIn URL, or Account Name plus First name and Last name - the import dialog was asking for a Domain column the import itself never required. Include it when you have it and the match to the right account is stronger; leave it out and nothing is blocked. See Import contacts from a CSV.

• Classic workflows now have a date. The older builder is being sunset at the end of 2026, replacing the earlier September date on the card. Workflows built there keep running until then; rebuild anything you rely on in the Workflow Builder before the deadline. See Workflow Builder and classic workflows.

• See what a launch will cost before you start it. Every agent launch dialog, and the Launch workflow dialog, now carries an Estimated N credits line beside the record count, re-pricing itself as you change the settings in front of you. Hover it for the breakdown by kind of work and the assumptions behind the figure. It is usually a range, because work your workspace has already done is not charged again and steps behind a qualification gate may never run. See Seeing the price before you launch.

• A run now reports what it actually spent. Open Run overview and the total credits the run has drawn sit in the header, with each step card carrying its own share, so an expensive run says which part of it was expensive. Per-record breakdown goes further and prices each individual record - usually the fastest way to see that a heavy batch was driven by a handful of records rather than by everything. See Monitor the task feed.

• Star an agent from wherever you are choosing one. The favourite star used to be clickable only on the agent lists. It is now live in the agent pickers inside the launch dialogs and the workflow builder, and on the research, account plan and play rows on an account or contact page. And a research agent you have favourited now gets its column in the default Accounts and Contacts view, instead of waiting for you to turn it on. See Mark an agent as a favourite.

• An email or phone found by the waterfall shows which provider found it. The icon beside the value is now that provider’s own brand mark rather than a generic one, so the source behind a number you are about to dial is visible at a glance. Providers we do not name individually keep the generic mark. See Find verified email and phone.

• A connected Slack or Teams card offers Manage instead of Add. Once the connection is live, the install button becomes a Manage menu keeping the reinstall link and the Teams app-package download reachable, the same shape a connected CRM card already had. See Integrations.

• Eva now answers in Microsoft Teams. The same colleague as in the workspace and in Slack: direct-message her, or mention her in a team channel or a group chat. She keeps a short work log while she works, shows what each step costs as it settles, and hands back the same tables, deep-link buttons and follow-up choices you get elsewhere. Your Teams administrator installs the app package once, downloadable from Integrations. See Talk to Eva in Microsoft Teams.

• Slack and Teams now have their own cards on the Integrations page. Each shows whether the connection is live, which chat workspace it is connected to, and when - so “are we connected?” is answerable without sending a test message. Slack installs straight from the card; the Teams card is where the app package is downloaded. See Integrations.

• You can hand Eva the exact thing you are looking at. Four new places open a conversation carrying the record with it, so she works on that account, contact or run rather than searching by name: Analyse with Eva on a run card in the task feed, Build with Eva in the Workflow Builder, Fix my ICP rules and Fix my IBP rules in the qualification status tooltips, and the existing Strategize with Eva on a record page. See Eva.

• Disagree with a qualification verdict without leaving the list. Hovering an ICP or IBP status now shows the reason, how long ago the record was qualified, and the actions that follow from the verdict: take the disagreement to Eva so the criteria or the vertical behind it get looked at, requalify that one record, or set the status by hand. See Record status fields.

• The ecosystem overview shows personas as a matrix. Persona cards run down the side, verticals across the top, and each cell says whether that card is assigned there - with a one-click Add persona to the vertical where it is not, and a Without persona card row so nothing is hidden. See Set up ecosystems and verticals.

• A new Has open task filter finds the accounts and contacts nobody is on. Set it to No and you get the records with no outstanding task, which includes the ones that never had one - the question is who needs picking up, not who once had work logged. It joins Has unanswered inbound and Has meeting material in the CRM activities group. See Accounts columns and filters.

• Eva’s work now draws credits, and you can see the price as it happens. What is charged is each piece of real work inside a turn - reading your accounts, searching for people, running an agent - not the conversation, the question, or the length of the answer. A price sits beside each step and a total under the finished answer, and nothing is charged for laying the answer out, for a step that failed, or for a step that found nothing, so the prices you can see are the whole cost. Her spend now has its own line in the credit usage report. See what a conversation with Eva costs and Read the credit usage report.

• Ask Eva for a list in Slack and she saves it as a view you can open. A list pasted into a Slack thread used to be a dead end; now the same slice arrives as a saved view on the Accounts or Contacts list with a button that opens it, named for what you asked. See Talk to Eva in Slack.

• A view Eva built no longer goes missing. Where her definition used a CRM condition the filter panel cannot compose - every part of it true of the same logged activity or the same deal - the view could fail to open and the list came back unfiltered. That condition is now kept intact and shown as a summary you can read and remove, with a note to ask Eva when it needs changing. See a condition Eva set for you.

• Multi-select filters got a Select all. In any filter section offering more than ten values, Select all appears beside Clear all and takes whatever the section is currently showing, so searching first and selecting the run is quicker than ticking each value. See Filter accounts and Filter contacts.

• CRM user mapping now tells invited and deactivated teammates apart. All three used to read as “no longer in the workspace”, which made a colleague who simply had not accepted their invitation look like a broken mapping. A pending invite now says so and needs nothing done; a deactivated Evergrowth user is flagged as one worth unlinking. See Connect a CRM.

• A lost invitation email is no longer a dead end. Forgot password? now works for a teammate who was invited but has never signed in, so they can set their own password without an admin issuing a reset link. See Sign in and reset your password.

• Eva now answers in Slack. The same colleague, in the channel where the question came up: mention her in a thread or send her a direct message and she works from the same accounts, contacts, personas and plays, at the same credit price as the web chat. A channel thread is shared - a colleague can carry on a conversation you started, and each person’s turn still runs with their own record access and their own allowance. She reads the replies your team added between her turns and picks them up the next time she is tagged, streams her working as she goes, and hands back the same tables, cards and follow-up choices you get in the workspace. A Slack workspace admin installs her once. See Talk to Eva in Slack.

• A contact keeps every email and phone number found, and you decide which one leads. Nothing is overwritten any more: the additional values sit behind the count beside the main one, and you can add or correct a value by hand, lock it so later finder runs cannot replace it, promote one of the others to main, or push a wrong one down so the next best takes over and a re-find cannot revive it. Where you have not decided, email leads on recency (an undeliverable address sinks below the rest) and phone leads on verification. That means a number a rep confirmed on a call survives every later enrichment run. See Find verified email and phone and the waterfall.

• A contact is no longer written to your CRM without its company. Accounts and contacts sync on separate tracks, so a contact was sometimes created in the CRM before the account it belongs to existed there. Those contacts are now held back with a status that says exactly that, and go through on their next attempt once the account lands. See My record will not sync to the CRM.

• A refused duplicate import is now visible. Uploading the same file while an identical import is still running was already blocked; the attempt now also appears in the Imports tab as a Duplicate card explaining why it was skipped, instead of leaving no trace. See My CSV import did not create the records I expected.

• The Accounts and Contacts lists now sort by almost any column. Clicking the arrow beside a column heading orders the whole filtered set by it: fit, Score, counts, dates, owner, email or phone coverage, profile strength, and the columns your own workspace defines - every qualification agent, research agent, play, and custom field. The first click leads with the strongest end of the column (biggest, newest, best-fitting) and a second click reverses it; Name still leads A to Z. Agent, play, and custom field columns order by when their value was last written rather than by its text, so reversing one gives you a work queue of the records whose answer is going stale, with the never-answered records at its head. The same sorting is on the child accounts and contacts tables of an account, and on a Workflow Scheduler automation’s assigned records. See Sort the Accounts and Contacts lists.

• Every agent answer and generated play on a list now carries a freshness clock. A small clock icon beside the value shows at a glance whether it is Fresh (under 14 days), Aging (14 to 30 days), or Stale (over 30 days), and hovering it gives the age in words. The windows are the same ones the freshness filters use, so a row and the toolbar always agree - filter to Stale, sort that column oldest-first, and work it down. See Accounts columns and filters and Contacts columns and filters.

• The CRM integration page is now four tabs instead of one long page. Each connected CRM opens on an Overview with the health of the connection (connection state, last inbound and outbound sync, the latest run’s results) and, until setup is finished, a checklist that counts off turning on your objects, mapping required fields, linking users, and running your first import. Objects takes one object at a time and draws it as a pipe between Evergrowth and your CRM, with a switch per direction and a filter control on each showing how many conditions are narrowing it; its field list is grouped, searchable, filterable by coverage, and can map every auto-matchable field in one pass. Users is now three sections - linked pairs, people in Evergrowth not linked yet (with a one-click email match), and CRM seats nobody is linked to - so the work left to do is visible. Activity splits into Sync runs (whole imports and exports, with a processed/rejected/in-progress meter per object) and Record log (record by record, with the full reason and payload behind each). Running a catch-up is now a single Run sync button offering an initial import or one custom object, and an explicit choice of whether the run may create records, update them, or both. See Connect a CRM.

• Quick actions on Home has been regrouped. The first three cards are now the sourcing and automation jobs, each ending in an Ask Eva action - find new accounts, find new contacts, build a workflow - and the fourth card, Improve your workspace, holds the setup and tuning work: connect your CRM, invite teammates, ownership, the “improve … with Eva” tuning actions, and duplicate cleanup. Connect your CRM disappears once a CRM is connected. See Find your way around the workspace.

• A finished run now announces itself on screen. When a workflow run completes, fails, or stops on credits while you have the workspace open, a card slides in at the corner with the outcome and its result chips, plus buttons into the per-record breakdown and the run overview. The Task feed control in the header also animates while anything is still queued or running, so in-flight work is visible from any screen. See Manage notifications and Monitor the task feed.

• Eva is clearer about long work and her own limits. A turn that takes real work now tells you how long it has been going and lets you stop it, and it keeps running if you leave the page - the conversation catches up when it is done. When she reaches the ceiling on how much she does in one turn she says what she covered and where she stopped, as information rather than an error, and a genuine failure is named (a step that timed out reads as a timeout) instead of being reported as a generic problem. She can also set up a recurring automation for you, confirming the cadence and per-run cost first, and tag the people in a segment as well as the companies. See Eva and Set up a Workflow Scheduler automation.

• Home opens with Quick actions. Four cards across the top of Home group the things you might be about to do by intent rather than by screen: Find new accounts, Find the right people, Put it on autopilot, and Set up your workspace. Hover a card and its actions drop open - add or import records, pull accounts in from LinkedIn or Crunchbase, clean up duplicates, build or schedule a workflow, connect your CRM, invite teammates. Several actions hand off to Eva to review the agents you already have and suggest the next one worth creating, and Find people at your accounts opens the Accounts list in a guided Contact finder state where your selection becomes the audience. Cards follow your permissions and hide rather than show half-empty. See Find your way around the workspace.

• More empty cells now offer the action that fills them, and the buttons say what they do. On top of the gaps already covered, an account with no Score offers research (narrowed to your signal research, since that is what the score counts), an account with no Contacts offers Find contacts, an account with no Account plans offers Generate plans, and a contact with no Plays offers Generate plays. Each button now carries its label in the cell instead of revealing it on hover, so a column of gaps is scannable. A contact that has never been searched for an email or phone shows just the finder button, with no “not searched” placeholder next to it. See Accounts list: columns and filters and Contacts list: columns and filters.

• A deleted ecosystem, vertical, or persona card is now something you can act on from the warning. The orange marker on a record whose qualification target was deleted used to tell you which advanced filter to build; it now offers Show all accounts like this (or contacts), which applies that filter for you, and Requalify, which re-runs qualification on the record. Qualification is also offered on the record’s own empty cells again, rather than being withheld while the marker was showing. See when a qualification target was deleted.

• An empty cell in the Accounts or Contacts list now offers the action that fills it. Instead of a blank placeholder, a gap shows the agent or action that closes it, launchable from the row without selecting it first: Domain finder agent on a missing domain, Qualification agent on an empty ICP, Ecosystem, or Vertical (and on an empty IBP, Type, or Expertise for contacts), Email/Phone finder on a missing email or phone with the right channel already chosen, Research agent on an empty research-agent column with that agent already selected, Generate plays on an empty play column with that play already checked, and Assign parent / Assign owner where those are unset. Empty custom-field cells keep their edit control visible rather than revealing it on hover. Each agent button only appears where the run could actually produce something - a qualification with no criteria behind it, or research by an agent that is not assigned to the record’s vertical or persona card, is not offered. See Accounts list: columns and filters and Contacts list: columns and filters.

2026-08

• Account research results now separate the reasons to act now from the facts you report on. The Research agents section of an account’s Research tab is split into two counted groups: Signals, the time-sensitive triggers that feed the strength score, and Data points, the standing account facts that sync back to your CRM. Each group says so in a line under its heading and has its own empty message, so a group with nothing in it reads as “that side of your research setup found nothing here” rather than as a quiet account. See Review agent outputs and The account detail page.

• Send a fixed value to a CRM field instead of the record’s own. A field mapping can now carry a constant - a lead source that always reads the same thing, a channel tag your reporting groups on - so you stop asking reps to type it onto every record. Set it under Always send a fixed value to your CRM on the outbound side of the mapping; leave it empty and nothing changes. Once a constant is set, you also choose whether it replaces the value already in your CRM or only fills the field while it is still empty, so a value a teammate typed by hand can be left alone. See Connect your CRM.

• Admins can publish a team template for every workspace. Make global template, on a team template’s card in the Workflow Builder gallery, moves that template out of Team templates and into Evergrowth templates. Steps that named particular verticals, agents, plans, or plays are widened to include all of them so each workspace fills them with its own data, and anything that cannot travel - a filter on your qualification answers, custom fields, or a named vertical or owner - blocks the move and is listed for you to clear in one pass. See Build a workflow.

• Credit limits: cap what a person or a team can spend. Credits are still a shared workspace pool, but an admin can now put a weekly and/or monthly ceiling on any person or any team, from the invite and edit windows or in bulk from Bulk update credit limits on the Users and Teams tabs. Nobody is capped by default. Everyone sees their own position in the header credits control, which now breaks down into Your limit, Team limit, and Workspace limit and shows whichever of the three will run out first. A run that hits a cap pauses with Limit reached, keeps the work it has finished, and names the cap that stopped it - and unlike running out of workspace credits, that one is not fixed by topping up: it needs a raised limit or the next window. See Credits and billing, Set credit limits, and A run stopped because a credit limit was reached.

• An empty relationship map now tells you why it is empty. The map only places contacts qualified as IBP yes, so it could read “you don’t have any contacts” while the Contacts tab showed dozens. It now names the actual blocker with the number behind it - nobody found yet, N contacts waiting to be qualified, or none of the N evaluated contacts matching your personas - and offers the action that clears it, including qualifying exactly those contacts or opening the ones that came back as a no. An empty expertise group does the same, and offers to look for that specific persona at that account. See Map account relationships.

• Launching a workflow asks the question plainly: all records, or stop at a goal. The launch window now offers Process all records and Stop early at a goal as two choices with the consequence spelled out for the selection in front of you, instead of a switch labelled “Stop when goal is reached”. A workflow saved with a goal opens on the goal, marked Workflow default, and changing it applies to that run only. Toggling between the two no longer discards what you typed. See Run a workflow on records now.

• Workflow Builder: drag steps onto the canvas, and see what a step needs before adding it. Steps in the Possible steps sidebar can now be dragged to a spot on the canvas, with a dashed outline showing where the card will land, and clicking one still adds it - now placed below the end of the flow rather than in a grid, with the canvas panning to it. Either way, a step added to a flow with a single loose end is wired to it automatically and says so. Hovering a step first shows when to use it, what it produces, and which prerequisite agents are already in this workflow. See Build a workflow.

• Account and contact pages: the left panel is gone, the details moved up top. Everything that used to sit in the resizable panel down the left side of a record - ecosystem and vertical, owner, parent, tags, email, phone, LinkedIn, relationship strength, languages, location, source - now sits in a block directly under the page header, above the tabs. Tabs get the full page width, the facts stay with you on every tab, and email and phone each get their own slot in a Reach row with their status, where the value came from, and the search and verify controls on the same line. Relationship strength is now a one-click menu instead of a modal, the ICP and IBP verdicts are now read and edited in one place - the header chip, with controls beside it to set the verdict by hand or lock it against the next qualification run - and a contact’s header shows the account’s strength score next to the account name. Custom fields left the panel too: they open in their own side panel from a button on the tab bar, with a count and, past five values, a search box. See Account detail page and Contact detail page.

• A channel that came back empty now offers alternatives, not just the search that failed. When the finder has already searched a contact’s email or phone and found nothing, that slot’s control opens a short menu instead: search again across all providers, reach out on LinkedIn, find someone else at the same account, or add the contact to Do Not Touch. See No email or phone found.

• Templates come to the Workflow Builder, and you can save your own. Add new on the Workflow Builder list now offers Start from scratch or Use template, opening the same gallery of Evergrowth and team templates the older builder had - previously the only way to start from a template was to build in the builder being phased out. The builder header also gains Save as template, so a sequence that works for your team can be saved once and reused by everyone. Each builder keeps its own set of templates, so a template saved in the older one needs rebuilding here. See Build a workflow.

• Pipeline insights diagnostics open the list behind the number. Each of the three checks now carries a Show the list link that opens the accounts or contacts it counted, filtered to exactly that cohort - the qualified accounts that did not score as a fit, the ready contacts nobody ever touched, the verified phone numbers nobody ever called. Accepting Eva’s offer on a check now also hands her that same cohort, its figures, and the period, so you no longer have to explain which records you mean. See Read the pipeline insights report.

• The task feed’s Agent steps filter reads like the rest of the workspace. Its options now use the agent names you see everywhere else - Account qualification, Account research, Contact qualification, Contact research - and structural steps such as a workflow’s routing decision are no longer offered, since filtering by them returned the run rather than the work. See Monitor running agents.

• Pipeline insights: what the agents produced, and what the team did with it. A new Pipeline insights page under Report counts supply and delivery side by side - accounts and contacts qualified, signals found and hours saved, then the same cohort followed forward to the point a rep logged an email, a call, or a meeting. It flags the reachable contacts nobody ever touched, splits outreach by channel, and runs three plain checks (list quality, activity volume, channel mix) that Eva turns into an offer of help rather than a verdict. The figures are counted from agent runs and CRM activity, refreshed nightly, with no model estimating anything. The same chain now sits as a collapsible banner above the Accounts and Contacts lists, and clicking a stage number filters the table to exactly those records. See Pipeline insights and Read the pipeline insights report.

• New filter: Has buying signals. Accounts and contacts can now be filtered by whether research answered any of your buying-signal criteria positively - the same rule the pipeline insights delivery chain uses, so the filter and the report agree. See Filter accounts and Filter contacts.

• CRM activity and opportunity filters now start at “at least 1”. Adding one of those groups in the advanced builder seeds the count condition ready to apply instead of leaving an empty condition you had to fill in first.

• The CRM sync entity is called Activities again, in the Run Sync window as well as on its tab, and the entity list no longer repeats an option or offers one your CRM cannot sync. See Connect your CRM.

• The task feed’s Action filter is now Agent steps, and it is available to everyone who can see the feed rather than to full-access members only. See Monitor running agents.

• A contract that has run out no longer counts as active. The credit balance treats a contract as in force up to and including its end date, so a lapsed term now shows the No active contract chip instead of reading as current. See Read the credit usage report.

• Report is now a credit usage report. Opening Report gives you Credit usage: your balance against its allocation, a weekly or monthly chart of spend stacked by agent across the last twelve periods, a ranked Spend by agent breakdown grouped into the same families as the agent library, and a Credit balance outlook projecting how the balance moves as your contracted grants land. Click any agent in the legend to isolate its trend, and read the Spend data updated line to know how fresh the figures are. The credits chip in the top bar also gained a View credit usage shortcut. This replaces the embedded analytics dashboard, which is still reachable from a link at the bottom of the page while it is retired. See Read the credit usage report.

• Account plans are priced per plan, not per account. Generating plans charges 1 credit for each plan produced, so selecting three Account plan agents for ten accounts writes thirty plans and draws thirty credits. The dialog’s footer count and the per-agent picker are where to check the scope before you start. See Generate account plans and Account Planning.

• Owners can now be assigned by rule instead of by hand. A new Ownership rules page under Settings lets you write prioritized rules - this filter to this team, that filter round-robin across those reps - and every account or contact that finishes qualification is routed automatically, with the owner pushed to your CRM as the native record owner. Rules run top to bottom and the first match wins, several assignees rotate evenly, teams pick up new members without the rule being edited, and a live preview shows exactly how your book would split before you switch anything on. Records matching no rule deliberately stay unowned rather than landing on a default owner, and any record can be locked so the rules never touch it. Contacts can simply follow their account’s owner, which is the default, or run their own rules where people at one company are worked by different reps. See Assign owners automatically.

• Imports no longer change the owner of records you already have. The old Override existing owners switch is gone. In its place, every surface that creates records - CSV imports, single adds, the data-source finder, the Chrome extension - asks who should own the records it creates: leave it to the ownership rules, take them yourself, or rotate them across chosen users and teams, optionally with an ownership lock so the rules never move them later. Owner-email columns still apply to created rows. Existing records keep their owners, full stop; reassigning those is now a deliberate act, either from the list or through the rules. See Import accounts and the CSV field reference.

• Your people can be mapped to their CRM seats. Each CRM integration now has a User mapping tab pairing Evergrowth users with CRM users, so an owner pushed to your CRM lands on the right seat even when the two email addresses differ. Unmapped people are matched by email automatically, suggested matches apply in one click, and a deactivated seat or departed teammate is flagged on the row instead of failing quietly. Owner changes made in your CRM now also flow back for mapped users. See Connect a CRM.

• Ownership is visible on the record. An account or contact’s Owned by field now says how it got its owner - a rule, an import, a person, Eva, or the extension - and carries a padlock that locks it against the rules. Lists show the padlock beside the owner and can be filtered by Ownership locked, and Edit ownership lock joins the bulk actions. See the accounts filter reference and Bulk-edit records.

• Activities and Opportunities can now be switched off per CRM connection. The two read-only feeds your CRM sends in - the logged calls, emails, meetings, tasks, and notes, and the deals - each have their own tab on the integration page with a single Sync status switch. Both start on, so nothing changes unless you want it to, and a workspace that would rather not carry that history can now stop it at the source instead of ignoring it. Accounts and Contacts have to be syncing first, since every activity and deal links to them, and switching a feed on governs what arrives from then on rather than importing history. See Connect a CRM.

• Whole blocks of workflow steps can be selected, copied, and pasted. Hold Cmd (Ctrl on Windows) and drag a box across the canvas, or Cmd-click cards one by one, then Cmd+C, Cmd+X, and Cmd+V. Dragging any card in the selection moves the group. The connections between the steps you copied are kept, so a qualify-research-plays block can be lifted into another workflow, or another browser tab, in one action rather than rebuilt step by step. See Build a workflow.

• The workflow canvas opens on the whole workflow. Reopening a workflow now fits it to the screen and centres it rather than starting at 100% in the corner, and you can double-click the background to zoom in on a spot (Cmd or Ctrl and double-click to zoom out). See Build a workflow.

• The permissions matrix reads as what it grants. The second column is now Manage rather than Edit, each row explains what its View and Manage actually allow when you hover it, and a column that would not change anything for a row shows a dash instead of a checkbox - so ticking a box always means something now. Widgets appears as its own group, with a separate permission for assigning widgets to other people. See Teams, roles, and permissions and Manage your team.

• The scheduler’s match count reflects the filter in front of you. Check match count in the Workflow Scheduler audience step now counts the audience as you have it on screen, unsaved edits included, instead of the version last saved - so you can size an audience while you are still shaping it. See Set up a Workflow Scheduler automation.

• An empty section on an account or contact now offers the agent that fills it. No plans, no research, no contacts, no plays, no persona card: each says what belongs there and gives you a button to launch the right agent against that record, so a gap is one click from being closed. See account and contact detail pages.

• You can now filter your lists by what has actually happened with an account, and by the pipeline on it. With a CRM connected, the advanced filter builder carries two new groups on both the Accounts and the Contacts list: CRM activities and CRM opportunities. Between them they cover how much is logged, when the last call, email, or meeting was, whether the last touch was yours or theirs, whose activity it was, whether a deal is open, what the won business adds up to, and the stage, amount, and close date of the deals themselves. On the Contacts list the same two groups also sit inside the Account group, so you can hold a persona-fit person against what is happening at their company. The options offered come from your own CRM, and an account’s history counts the records logged against its people as well as its own. This is what makes lists like “fits our profile and already in conversation”, “was active last quarter and has gone quiet”, or “reachable contacts at accounts with a live deal” a saved view rather than a manual exercise. See Filter accounts, Filter contacts, and the accounts and contacts filter references.

• Columns can be reordered straight from the table. Hover a column heading, take hold of the grip at its left edge, and drag it where you want it; the other columns move aside as you go. The first column stays pinned. Opening the column picker just to change the order is no longer necessary. See Choose and reorder table columns.

• A workflow step can be duplicated instead of rebuilt. Each step card in the workflow builder now has a menu with Duplicate and Delete. Duplicate drops a copy with the same settings next to the original, ready for you to connect and adjust - useful when two steps differ in one setting. Deleting a step now also tidies up the connections that pointed at it. See Build a workflow.

• A contact’s DISC read is now shown as a wheel. The DISC profile leads with a four-part wheel with this person’s traits highlighted, on the contact’s Research tab and next to the suggested tone of voice on Overview, so you can see at a glance how to pitch to them before reading the written profile. See Open and read a contact detail page.

• Accounts and contacts now show how old their qualification is. The record panel carries a Qualified at value written as an age, with the exact date on hover, so a verdict from months ago reads as one instead of looking current. See Open and read an account detail page and Open and read a contact detail page.

• The account and contact pages have been rebuilt around a panel that always shows the record. Both pages now carry the same shell: a column down the left holding who the contact is or what the company is - contact details and persona mapping, or company information - plus that record’s custom fields, visible whichever tab you are reading and resizable by dragging its edge (the width you pick is remembered). The long strip of tabs is gone. A contact has four - Overview, Research, Practice, Activity - with LinkedIn insights, the DISC profile, and the persona card folded into Research, the digital twin and roleplay under Practice, and the task feed, CRM records, and changelog under Activity. An account has Overview (its account plans), Contacts, Research (which now also holds the business model), Relationship map, Activity, and Child accounts. Links you saved to the old tabs still open the right place. See Open and read an account detail page and Open and read a contact detail page.

• Eva can now build your account list, not just work the accounts you have. Describe the companies you are after and she searches the LinkedIn and Crunchbase company data, tells you how many match and how many are genuinely new to you, and brings them in as accounts once you confirm the number. Companies already in your list are left untouched, and the accounts land on their own with you as the owner. See Add accounts from LinkedIn or Crunchbase and Eva.

• The IBP column now sits near the front of the contacts list. It has moved to just after Account Score, so whether a contact is an Ideal Buyer Profile match is visible without scrolling sideways. See Contacts columns and filters.

• An empty table now says so where you are looking. When a filter or search returns nothing, the empty-state message replaces the table instead of being drawn inside it, so on a wide list it is no longer pushed off to the side where you have to scroll to find it.

• Ecosystems and research agent sandboxes are now two separate pages. The Agent Training Center overview offers a card for each, so you open the one you want instead of landing on a shared page and switching tabs. Ecosystem cards now show their verticals, their persona and vertical counts, and when they were last updated, with ongoing ecosystems listed first, and you can switch an ecosystem between Ongoing and Paused straight from the chip on its card. A sandbox holds accounts rather than verticals and is marked as a test environment. Adding or editing a vertical (or a sandbox account) now walks through a short series of steps, with a step each for the qualification and research agents that apply to it, so it can arrive fully configured rather than empty; duplicating one opens the same steps prefilled from the original. See Set up ecosystems and verticals and Ecosystem and vertical fields.

• An agent can now be added to an ecosystem without building it from scratch, and moved between a sandbox and a live market. The plus icon on a Qualification Agents or Research Agents group offers Create new, Add agent used in other ecosystems, and Add from template, and a research group adds Add agent from sandbox so something you proved on a few named companies can be promoted into a live ecosystem. Inside a sandbox the same menu offers Add agent from verticals, the other direction, so a live agent can be reworked in the sandbox without disturbing the version your reps depend on. See Set up ecosystems and verticals and Create qualification and research criteria.

• Eva now acts on a sensible reading instead of asking what you meant. A request like “give me my top ten accounts” is answered with the accounts, ranked, plus a line naming the measure she ranked them by and the other readings offered as choices you can select. Answers arrive in priority order with a stated first move, records come back laid out to scan or open rather than typed into a paragraph, and when she cannot do something she names the limit in one sentence and delivers the manual route with the content ready to use in the same reply. See Eva.

• Eva can see which CRM you have connected. She reads your workspace’s own connections, so she works from your CRM data instead of telling a connected customer to connect a CRM or declining work she can do. If that check fails on a given turn she says she could not confirm it and tries anyway rather than reporting that nothing is connected, and she no longer reports an empty set of persona cards for a workspace that has them. See Eva and Integrations.

• Dropdown options from your CRM stay current on their own. When your team adds an option to a dropdown field in the CRM after its mapping was saved, incoming records used to arrive with that field empty while still reporting as synced. Evergrowth now refreshes the option lists for mapped dropdown fields every hour, and again the moment a record arrives with an option it does not recognize, so the value resolves from then on. Records that already came in blank pick the value up the next time they change in your CRM. See Connect your CRM.

• A sequence play now comes back with every step you asked for. A play built with more than three steps used to risk having its later steps merged together; each step is now written on its own, so a six-step sequence returns six distinct messages. Steps also keep their positions: if one produces no copy it stays as an empty step at its own number instead of dropping out and pulling the rest up, so a gap reads as an incomplete run rather than a shorter play. See Generate plays for a contact and A step in a sequence came back empty.

• Tag changes now show in an account’s or contact’s Changelog. Adding or removing a tag is recorded like any other change, with who or what did it and when, so tags are attributable rather than just different from what you remember. At the same time the Changelog stopped logging entries where nothing actually changed, so a routine re-sync that rewrites the same values no longer pads the history. See Open and read an account detail page and Open and read a contact detail page.

• Value proposition sources are now matched by their plain domain. A website added with a prefix is listed and looked up under the shortest form of its address, so it stays one source however you type it, and generation finds its content instead of waiting on a source it cannot match. When a source cannot be used, generation now names it and says whether to wait or to remove and re-add it. See Write and publish your value proposition.

• A play can now see what your colleagues discussed at the same account. When a play’s instructions ask for prior interactions, the Play Copywriting agent now also gets a short read of the calls and meetings your team logged with other people at that account - a routing instruction like “go through the VP, not the analyst”, an objection someone else raised, a commitment your team made, or where the deal stands. It arrives with the same request rather than needing a separate one, and it is the piece that was previously invisible when writing to a new person at an account you already work. Each interaction now also records who it was with, so nothing a colleague said is attributed to the person you are writing to. See Play Copywriting and Write effective play instructions.

• Qualification and research answers can now be edited straight from the ecosystem table. Hover the cell where a criterion row meets a vertical column and a pencil opens a small editor over that cell: set or change the answer, save with the checkmark, discard with the cross. Setting an answer where there was none also switches that vertical on for the agent, and the trash icon unassigns it again. Research Agent rows now show at a glance whether the agent is Enabled or Disabled for each vertical, with the pencil offering Assign or Unassign. The pencil on the criterion row itself opens the full agent editor, for changes to the question rather than the answer. See Create qualification and research criteria and Set up ecosystems and verticals.

• Account plans can now be written from your CRM history. When your CRM is connected, the Account Planning agent can draw on what your team has already logged with an account - emails, calls, meetings and notes across its people, plus the deals open or closed against it - so a plan reasons about where the relationship actually stands instead of treating a worked account as new. The history is gathered across the account’s people rather than from the busiest thread, your logged calls and meetings are always kept, and an account with nothing on record is planned as a genuine first approach. See Account Planning and Generate account plans.

• Uploading the same import file twice is now refused instead of hanging. Submitting the identical CSV again while the first import is still running is rejected up front with a message saying an import of that exact file is already in progress. Previously the second upload appeared to queue and then sat there indefinitely. See My CSV import did not create the records I expected.

• Contacts now show their profile photo. Where a photo was found while enriching a contact from LinkedIn, it appears on the contact’s avatar in the contacts list and on their detail page; initials are still used when there is none. See Open and read a contact detail page.

• Account agents can now be launched from the Contacts list. The agent picker on the Contacts list has an Accounts tab alongside the contact agents. Pick an account agent there and the workspace resolves your selected contacts to the companies they belong to - duplicates collapse to one account - then runs against those. Start from the people you care about and work their companies without rebuilding the same selection on the Accounts list. See Run a Qualification agent, Run a Research agent, Generate account plans, Run the Contact Finder, and Run the Domain Finder.

• Plays can now be written from your CRM interaction history. When your CRM is connected and a play’s instructions ask for it, the Play Copywriting agent can draw on what your team already logged with that contact - past emails, calls, meetings and notes, plus the deals open or closed on their account - so a message can build on a real conversation instead of opening cold. It references only what is actually logged, and a contact with no history is written to as a genuine first touch. See Play Copywriting and Write effective play instructions.

• Admins can generate a password reset link and share it directly. In Settings under Users management, the Reset password action on a person’s row now produces a copyable link instead of sending an email, for cases where the emailed reset is not reaching them. The link is valid for 7 days, and generating a new one invalidates older ones. See Manage your team.

• The forgot-password screen now confirms where the link went. After requesting a reset you see the address it was sent to, can resend after a short countdown, and can switch to a different email without starting over. Only the newest link works. See Sign in.

• A play that fails technically now shows as Failed. Previously a play that hit an error while generating could arrive as generated-but-blank; it now carries a Failed status with a technical-error reason, so a run that did not complete is distinguishable from a play that had nothing to say. See A play produces poor or inconsistent output.

• Eva now keeps a history of your conversations. Open an earlier conversation to pick up where you left off, or start a new one for a different question - your past conversations stay available instead of disappearing when you close the chat. See Eva.

2026-07

• A run that stops for credits can now be resumed. If a task feed run stops partway through because your company ran out of credits, it now shows an Out of credits status instead of failing outright. A Continue workflow button (on the task card, in Run overview, and in Per-record breakdown) picks the run back up from where it paused once your workspace has credits again, without redoing records that already finished. See Monitor running agents in the task feed.

• CRM periodic sync now includes Engagements and Opportunities. Previously these calls, emails, meetings, and deals only came in through a manual Run Sync; now they arrive automatically on your periodic sync frequency, the same as Accounts and Contacts, wherever your CRM supports them. See Connect a CRM.

• A task’s run report is now two views. Choose Run overview for the flow-map of the whole run (previously called See Report), or Per-record breakdown to search for one account or contact in the run and see just its own path through the steps, including which way it went at any branch. This per-record view is now open to any team member who can launch workflows, not only Admins. A step with skipped or errored records also shows an expandable list of the specific reasons, each with how many records it applied to. See Monitor running agents in the task feed.

• CRM sync now labels calls/emails/meetings as Engagements. In the integration’s Run Sync window, sync history, and sync logs, this record type is now labeled Engagements instead of Activities, matching the term used elsewhere in the sync setup. This is a label change only, not a change to what syncs; the Activities tab on account and contact detail pages is unaffected. See Connect a CRM.

• Sign in with your Google account. The login screen now offers Continue with Google next to the usual email-and-password login, so you can sign in with a Google popup instead of typing credentials. Not available inside the Chrome extension. See Sign in.

• CRM meeting summaries now show on the Activities tab. When your CRM captures an AI-generated meeting summary or internal notes for a call or meeting, it now appears in an expandable Internal notes section on that activity, and the activity also lists the contacts involved. See Open and read an account detail page and Open and read a contact detail page.

• Manage columns redesigned. The column picker now shows both available and selected columns as chips you click or drag, with available columns grouped by type (agents, plays, custom fields, and so on) when there is more than one kind. See Choose and reorder table columns.

• Save-as-default and reset moved into one menu. On the Accounts and Contacts lists, Save view as default, Reset to default view, and Reset column widths now live together in the overflow menu (…) next to Manage columns, instead of as separate toolbar buttons. See Filter and search the accounts list and Filter and search the contacts list.

• Edit custom field values directly from the column. In the Accounts and Contacts lists, a custom field column is now click-to-edit - update the value in place without opening the record. Clearing it removes the field’s assignment for that record. See Define custom fields.

• Filter and choose columns on the Manage Duplicates views. The account and contact duplicates views now support the same advanced filters, saved views, and search as the main Accounts and Contacts lists, on top of the existing search/status/origin filters - useful for cleaning up duplicates within one ecosystem or owner at a time. These filters reset when you leave the page. On accounts, auto-merge also respects any filters applied when you open it, and only merges the matching groups. See Resolve duplicate accounts and Resolve duplicate contacts.

• Edit account and contact details right on the page. The account and contact detail pages now let you edit the name, domain, LinkedIn URL, job title, languages, and location in place - hover the field, click the edit icon, and it saves as soon as you confirm it. There is no separate edit window for these fields anymore. See Open and read an account detail page and Open and read a contact detail page.

• One “Launch agent” button instead of a row of separate ones. The account and contact detail pages now offer a single Launch agent button that lists every agent you can run against that record - each with a short description and when to use it - instead of a separate button per agent. See Open and read an account detail page and Open and read a contact detail page.

• Strength score now shows as a ring, not a chip. The account strength score - on the account detail page, in list columns, and on a contact’s Account Score - now displays as a colored ring (around the company logo on the detail page, or with the number inside it elsewhere) instead of a plain colored chip. The color bands and meaning are unchanged. See Record status fields.

• Choose which columns show on the duplicates pages. The Manage Duplicates views for accounts and contacts now let you pick and reorder the columns shown inside each group’s table, the same way you would on the main Accounts or Contacts list. See Resolve duplicate accounts and Resolve duplicate contacts.

• Generate your value proposition from your website or documents. The Value Proposition page now has a Sources panel (Admin team members only) where you add PDF documents or website URLs and let the agent draft sections from them - either rebuilding the whole draft or adding new sections alongside what you already have. The publish confirmation was also redesigned to show exactly which sections changed and a “will sync” summary of the ecosystems, verticals, and persona cards the update reaches. See Write and publish your value proposition.

• Fewer places a criterion’s answer can feed into. When creating or editing a qualification or research criterion, the reuse options are down to two: the account’s business model summary or Signals - the previous options to route an answer into account plans or the competitive landscape view are gone. See Create qualification and research criteria and Criterion configuration options.

• See where a contact’s email or phone came from. Contact cards, the contacts list, and the contact detail page now show a small icon next to a found email or phone number - hover it to see whether it was synced from your CRM, added through a CSV import, or found by the Email/Phone finder (and when). A contact that has never been searched now reads Not searched instead of a blank field. See Record status fields.

• Choose whether the Email/Phone finder re-searches contacts that already have a value. When you turn on email or phone search - in the manual Find emails & phone numbers action or the Email/Phone Finder workflow step - you can now pick Keep the existing email/number (the default, which skips a contact that already has one) or Always search for a fresh one (searches anyway and can replace what is there). Previously a contact with an existing value was always skipped. See Find verified email and phone, and verify phones and Workflow step types and settings.

• Contact finder redesigned as a two-step flow. Finding contacts for accounts now walks you through Contacts to find (who and how many) and then Search settings (effort, email, and phone), and the email/phone “required” toggle is now a clearer choice: Skip and try someone else or Add them anyway when a contact has no email or phone. The per-persona counts also have a new roster view with a running total and a plus/minus counter per persona. See Find contacts for an account (Contact finder) and Workflow step types and settings.

• Choose how many records a velocity widget launch covers. Launching a widget’s workflow now opens a popover where you pick how many of the matching records to include, from one up to the full count, instead of always running against everyone that matches. See Set up and use velocity widgets.

• Merging accounts or contacts no longer proceeds locally when your CRM rejects the merge. If a connected CRM rejects the merge for one or more selected records, those records are left as they are instead of being merged in the workspace only - this stops the same duplicate from being re-created by your next CRM sync. A manual merge shows an error naming what was held back; an auto-merge run that hits this finishes with an “Auto-merge finished with issues” notification. See Resolve duplicate accounts and Resolve duplicate contacts.

• Richer call and meeting activity on account and contact pages. The Activities timeline on the CRM tab now shows each call or meeting’s outcome, direction, and duration, plus a link to the recording and an expandable transcript when your CRM captured them. See account detail pages and contact detail pages.

• Velocity widgets: new dashboard tiles pairing a working set with one-click launch. Your Home dashboard now has a Your universe section where you can create live tiles that track a working set of accounts or contacts (from a saved view or a filter you build) and launch an agent or workflow against whoever currently matches, in one click. Widgets can be personal or shared with your team, start from a gallery of ready-made templates, and refresh their counts on their own. See Velocity widgets and Set up and use velocity widgets.

• Key Accounts has been removed. The Key Accounts ecosystem type, the dedicated Key Accounts page, and the “Vertical / Key account” labeling across accounts and contacts have been removed. Ecosystems now come in two kinds: standard ecosystems and Research Agent Sandboxes. See Set up ecosystems and verticals.

• Workspace credits popover redesigned. The header credits chip and its popover now lead with credits remaining rather than credits used, adding a % left indicator and a separate line for credits used this period. See Top up credits and manage billing.

• The industry glossary has been removed. The Agent Training Center’s “Industry jargons” area, where teams defined market-specific terms for agents to reference, is no longer part of the product. Its how-to page has been removed along with related links. See The Agent Training Center.

• The Digital Twin is generated automatically, with no separate step. Qualifying a contact as a fit for your ideal buyer profile now generates its Digital Twin on its own - there is no longer a manual “Generate digital twin” action, and no separate Digital Twin step to add in the Workflow Builder. See Digital twin and Workflow step types.

• The sidebar now starts expanded. The first time you sign in, the left navigation opens with labels showing instead of collapsed to icons, so new users can see what each icon means. Once you collapse or expand it yourself, your choice is remembered as before. See Find your way around the workspace.

• Multi-word account search. Searching the accounts list by name now matches each word you type separately, so a search like “Orora Tech” also finds an account named “OroraTech”. See Filter and search the accounts list.

• Credits chip shows abbreviated totals. The header credits chip now abbreviates larger numbers to thousands (K), millions (M), or billions (B), for example 1K / 5K, keeping the exact count below 1,000. See Top up credits and manage billing.

• The Workflow Scheduler replaces Scheduled Tasks. Automating a workflow now covers a lot more ground: point it at a fixed set of records, an existing saved view, or a custom filter you build on the spot; optionally give it a goal and a safety cap on how many records one run can touch; and choose between a daily/weekly/monthly cadence or a new realtime trigger that fires the moment a record starts matching, instead of waiting for the next scheduled batch. Every run now shows up in a Run history on the automation’s page, linking straight to the task feed. See Set up a Workflow Scheduler automation and Workflows.

• The task feed now defaults to your own tasks. The header Task feed panel and the full Agent tasks feed page show only the tasks you started, so your view stays focused on your own work. Change the Created by filter to see a teammate’s tasks too. See Monitor the task feed.

• A dedicated Workflow Builder, and a heads-up on classic workflows. The workflow area now opens with a hub showing two ways to build a custom workflow: the current Workflow Builder (branching, goals, the one described throughout these docs) and the original builder, now labelled Classic and being phased out. If you still rely on a classic workflow, rebuild it in the Workflow Builder before Classic is retired. See Workflows and Build a workflow.

• A richer run report. The task feed’s See Report view now shows each step’s counts as a row of icons instead of a text summary, lets you switch between All, New this run, and Reused results, and breaks down why contacts were disqualified from persona qualification. See Monitor the task feed.

• Delete accounts or contacts straight from Manage Duplicates. The duplicates view now has a Delete action alongside Merge and Not a duplicate, so you do not have to leave the view to remove records you do not want to keep. See Resolve duplicate accounts and Resolve duplicate contacts.

• See an account’s or contact’s CRM pipeline and history in one place. Account and contact detail pages now have a CRM tab that surfaces the records synced from your connected CRM, so a rep has the deals and the engagement history right next to the research. It has three sub-tabs: Opportunities (the deals, with value, stage, expected close date, and whether each is Open, Won, or Lost), Activities (a timeline of logged calls, emails, meetings, tasks, and notes, with upcoming items flagged), and Logs (the most recent sync events for the record). The tab appears on any record linked to a CRM, and everything in it is read-only. See account detail pages and contact detail pages.

• Scheduled tasks are managed in one place. The per-record Scheduled tab has been removed from account and contact detail pages. Manage every scheduled task from Automations, Scheduled Tasks; you still schedule a workflow for a record the same way, from the Select workflow action. See Schedule a recurring workflow.

• A clearer report on every agent run. Choose See Report on any task in the task feed to see the run drawn out as a visual flow: each step is a card showing how many records completed, were skipped, or errored and how long it took, branches are labelled with how many records took each path, and clicking a step opens its detailed outcomes and the settings it ran with. Drag and zoom to move around a large run, and watch the counts fill in live while it is still going.

• Start a workflow from a template. When you build a workflow you can now begin from a ready-made template instead of a blank canvas - either one Evergrowth provides for a common motion or one your team has saved - then adapt it and make it your own. See Build a workflow.

• Sync deals and activities from your CRM. CRM sync can now read in two more kinds of record so your agents work with the full context of an account: deals (opportunities) and activities (the calls, emails, meetings, notes, and tasks logged in your CRM). Deals and activities both sync from every supported CRM; both are read-only, flowing from the CRM into Evergrowth so research and qualification can see the pipeline and engagement history already on an account. Choose them when you connect or sync a CRM, and see Integrations for how they fit.

• See your next credit grant. The credits indicator and the Report now show when your next credit grant arrives and how many credits it will add, so a refresh is something you can plan around. See Credits & billing.

2026-06

• Teams, roles, and granular permissions. You can now control exactly what each person can see and change. Team management has moved to Settings, Users management, where you manage Teams - each carrying its own permissions across the workspace - and Users, the people you invite and assign to a team. Every workspace starts with four ready-made teams - Admin, Sales Ops, Sales Rep, and Read-only - and you can rename them, adjust their permissions, or build your own. See Manage your team and Teams, roles, and permissions.
• Filter duplicate accounts by origin. The Manage Duplicates view has a new Origin filter, so you can focus on duplicate groups that came entirely from your CRM, entirely from Evergrowth, or a mix of both - useful for reconciling records that exist in both places before a sync.
• First full documentation release. Complete coverage of the Evergrowth workspace: the Agent Training Center and its four pillars, every agent, every playbook, how to work day to day (accounts and contacts, running agents, credits, integrations), and the thinking behind it (context vs data, how Evergrowth compares).

Eva note: this page is the human-readable record of doc changes. When a customer asks “what changed” or “is this new”, summarise the most recent entries here and link the affected pages. Entries are added as documentation is regenerated; an empty or sparse month means no customer-facing doc changes landed that month, not that the product stood still.