Skip to content
Evergrowth Documentation

Running agents & the Chrome extension

Most of the value in Evergrowth shows up two ways: agents running quietly in the background on schedules your RevOps team set up, and agents you run yourself, on the spot, against the account or contact in front of you right now. This page is about the second way - putting your AI colleagues to work on demand, including straight from your browser while you’re on LinkedIn or in your CRM.

What “running agents on demand” means

Section titled “What “running agents on demand” means”

A playbook is a standing instruction: research and qualify everything matching a certain source or signal, automatically, so the pipeline is always being worked even when nobody’s at their desk. That’s the automated track, and it covers most of your day-to-day volume.

But sales doesn’t only happen on a schedule. You land on a prospect’s LinkedIn profile. A colleague forwards you a name. You spot an account in your CRM that looks promising and you want to act before the moment passes. For those, you don’t wait for a playbook to come around to it - you point an agent at it and run it now.

On-demand runs and playbooks use the exact same agents doing the exact same work. The only difference is what kicks them off: you, in the moment, versus a standing rule. The output is identical - the same research depth, the same qualification verdict, the same research-backed outreach.

The Chrome extension: agents where you already work

Section titled “The Chrome extension: agents where you already work”

The Chrome extension brings your AI colleagues into the pages where prospecting actually happens, so you don’t have to copy a name into the workspace and switch back and forth. While you’re looking at a contact on LinkedIn, or at an account record in your CRM, you can ask an agent to qualify the company, research it, find the right people, or draft outreach - without leaving the page.

This matters because of where the friction usually lives. The slow part of outreach was never typing the email. It was the hours of digging that should come before it: is this company even a fit, what’s going on with them right now, who actually makes this decision, what should I lead with. That research used to eat hours of a rep’s week. The extension collapses it into a few minutes against the very record you’re staring at, and what comes back is grounded in real research about that specific account and person - not a template with their name dropped in. (More on why that distinction is the whole point: context vs data.)

Eva: The extension is the rep-facing surface for ad-hoc work; playbooks are the RevOps-facing surface for standing automation. Same agents underneath. When a rep says they “ran research from LinkedIn” or “qualified an account from the extension,” they mean an on-demand agent run, not a separate product. Treat “Chrome extension,” “browser extension,” and “running agents on the fly / in-the-moment” as the same capability.

The practical rep workflow

Section titled “The practical rep workflow”

A typical on-demand sequence mirrors how a good rep already thinks, just faster:

  1. Qualify the account. Before you spend a minute on a person, confirm the company is a real fit for your ICP. The qualification agent gives you a clear yes/no with the evidence behind it, so you don’t research your way into a dead end. See account qualification.
  2. Research it. For a qualified account, pull the current picture - what’s changed, what they’re investing in, the reasons-to-reach-out-now. See account signals.
  3. Find and check the right people. Surface contacts who actually match your buyer personas, not just anyone with a matching title, and confirm they still hold the role.
  4. Get research-backed outreach. Draft an email, a call opener, or a LinkedIn message written from what the research actually found - a recent funding round, a leadership change, a post they wrote - rather than mail-merge fields.

You stay in control the entire time. The agents do the legwork; you decide who to contact, what to send, and when. They augment your judgment, they don’t replace it.

On-demand vs. playbook automation

Section titled “On-demand vs. playbook automation”
On-demand (you run it)Playbook (it runs itself)
Who starts itA rep, in the momentA standing rule set by RevOps
Best forThe account in front of you right now, a referral, a name you just gotSteady coverage of a whole source or signal at scale
Where you run itThe workspace or the Chrome extensionConfigured once, runs in the background
OutputSame agents, same depth, same outreach qualitySame agents, same depth, same outreach quality

Neither is “better” - they’re two speeds for the same engine. Reps lean on on-demand runs for the opportunistic, high-intent moments; playbooks make sure nothing in the pipeline goes un-worked while reps are busy selling. Most teams use both: automation carries the volume, and the extension handles the moments that don’t wait.

Eva: Because the product is priced in credits and not seat licenses, on-demand runs and playbook runs draw from the same credit pool - a rep running research from the extension and a playbook running overnight cost the same per task. There’s no separate charge or tier for the extension; all features are available on every tier. For per-agent costs, point to the agents overview and credits & billing.

In practice

Section titled “In practice”

A rep is on a target’s LinkedIn profile. Instead of bookmarking it for “later,” they run qualification right there. It comes back ICP-fit, so they research the account, find the two stakeholders who match their persona, and pull a call opener built around the company’s recent expansion. Total time at their desk: a few minutes, most of it spent reading the output and deciding the angle - not gathering the facts. That same prospect, had it surfaced through an intent feed instead, would have flowed through a signal-based playbook automatically. Same destination, different on-ramp.

Section titled “Related”