This is the full catalogue of steps the workflow builder offers and the settings each one exposes. Use it when you are assembling a multi-step run and want to know exactly what every step does and every toggle it carries. The builder groups its steps into two lists: Agents (the work a step performs on a record) and Routing (the Condition step that sends records down different paths).
The Workflows builder is an internal, environment-limited surface: it is available only to administrators and only outside production, so most teams will not see it in the left navigation.
Each agent step runs one agent against every record that reaches it. You pick the step from the Possible steps list on the left of the builder, drop it on the canvas, and open it to set its options.
| Step | What it does |
|---|
| Domain Finder Agent | Finds and verifies the company’s website. |
| Account Qualification Agent | Checks whether the account fits your ICP. |
| Account Research Agent | Researches accounts for relevant signals and context. |
| Account Plans Agent | Turns account research into a strategic plan. |
| Contact Finder | Finds the right people at each account. |
| Contact Qualification Agent | Checks whether a contact matches your buyer persona. |
| Contact Research Agent | Researches contacts for relevant context. |
| Digital Twin Agent | Generates a digital twin of a contact to enable Copilot, Roleplay, and play personalization. |
| Email/Phone Finder | Finds a verified email and phone for a contact. |
| Play Agent | Writes personalized outreach for a contact. |
Some agent steps depend on earlier ones. When a step needs prior results, the builder shows a note at the top of its panel listing what the record must already have. Those prerequisites are:
| Step | Record must already have results from |
|---|
| Account Research Agent | Account Qualification Agent |
| Account Plans Agent | Account Qualification Agent |
| Contact Finder | Account Qualification Agent |
| Contact Qualification Agent | Account Qualification Agent |
| Contact Research Agent | Contact Qualification Agent |
| Digital Twin Agent | Contact Qualification Agent |
| Play Agent | Account Qualification Agent, Account Research Agent, Contact Qualification Agent |
Domain Finder Agent and Email/Phone Finder have no prerequisites.
Every agent step opens with a switch that lets you reuse earlier results instead of running the agent again. It saves credits and keeps a workflow consistent with previous runs. The wording is tailored to each step:
| Step | Switch heading |
|---|
| Domain Finder Agent | Skip accounts that have already been searched |
| Account Qualification Agent | Skip accounts that have already been qualified |
| Account Research Agent | Skip accounts that have already been researched |
| Account Plans Agent | Skip accounts that already have an account plan generated |
| Contact Qualification Agent | Skip contacts that have already been qualified |
| Contact Research Agent | Skip contacts that have already been researched |
| Digital Twin Agent | Skip contacts that already have a digital twin |
| Email/Phone Finder | Skip contacts that already have a verified email and phone |
| Play Agent | Skip contacts that already have plays generated |
Contact Finder does not carry this switch.
The agent steps that draw on a library of items you configure in the Training Center (research agents, account plans, plays) and the Account Qualification step (which draws on your verticals) share an Auto-include all switch:
| UI label | What it means | Values |
|---|
| Auto-include all | Selects every item and automatically includes any new ones added in the future. | On / off |
When Auto-include all is off, you pick the exact items from a searchable list. What you pick differs by step:
| Step | What you select when Auto-include all is off | Validation |
|---|
| Account Qualification Agent | The verticals (grouped under their ecosystems) to qualify against. | At least one vertical required. |
| Account Research Agent | The account research agents to run. | At least one required. |
| Account Plans Agent | The account plans to generate. | At least one required. |
| Contact Research Agent | The contact research agents to run. | At least one required. |
| Play Agent | The plays to generate. | At least one required. |
Filtering records to a subset of the accounts or contacts in the run (rather than choosing which agents fire) is done with a Condition step, described below.
| UI label | What it means | Values |
|---|
| Skip accounts that have already been searched | Reuses an earlier Domain Finder result for the account instead of searching again. | On / off |
| Effort level | How hard the agent works to identify and confirm the company. Low Effort is best when the right company is easy to find or confirm; High Effort is best when identifying the correct company needs deeper reasoning. | Low Effort (1 credit), High Effort (3 credits) |
| UI label | What it means | Values |
|---|
| Skip accounts that have already been qualified | Reuses the previous ICP decision instead of re-qualifying the account. | On / off |
| Auto-include all | Qualifies against every vertical and auto-includes new ones. | On / off |
| Verticals | When Auto-include all is off, the verticals to qualify against, grouped by ecosystem and searchable. | Your configured verticals; at least one required |
| UI label | What it means | Values |
|---|
| Skip accounts that have already been researched | Reuses existing research findings instead of researching again. | On / off |
| Auto-include all | Runs every account research agent and auto-includes new ones. | On / off |
| Agents | When Auto-include all is off, the account research agents to run. | Your configured research agents; at least one required |
| UI label | What it means | Values |
|---|
| Skip accounts that already have an account plan generated | Keeps an existing plan instead of generating a new one. | On / off |
| Auto-include all | Generates every account plan and auto-includes new ones. | On / off |
| Plans | When Auto-include all is off, the account plans to generate. | Your configured plans; at least one required |
Contact Finder combines finding the right people with optionally searching for their email and phone. It has no skip-already-processed switch.
| UI label | What it means | Values |
|---|
| Search for email | Turns on email finding for each contact found. | On / off |
| Email is required | When on, the system looks for an additional contact if no email is found. Contacts without an email are still added to your list but are excluded from later steps (Contact Research, play generation) so you don’t spend credits on contacts you can’t reach. | On / off |
| Email verification | Verifies found emails with third-party providers. | On / off |
| UI label | What it means | Values |
|---|
| Search for phone number | Turns on phone finding for each contact found. | On / off |
| Phone number is required | When on, the system looks for an additional contact if no phone is found. Contacts without a phone are still added to your list but are excluded from later steps (Contact Research, play generation). | On / off |
| Phone intent scoring | Scores phone numbers for buying intent. Available only when the phone-scoring integration is connected; otherwise the option is disabled with a link to set it up. | On / off |
| Phone scoring model | Shown when phone intent scoring is on. Lightning Scoring is a single-pass model that identifies a large portion of high-propensity contacts (under-24-hour turnaround); Titan Scoring uses a multi-pass model to extract every identifiable high-propensity contact (under-5-day turnaround). | Lightning Scoring, Titan Scoring |
You must enable Search for email or Search for phone number (or both) for the step to be valid.
| UI label | What it means | Values |
|---|
| Effort level | How much reasoning the agent applies to each candidate. Low: best when your persona is simple and clearly defined; the same contact search as Medium with less per-match assessment. Medium: looks more carefully at each candidate before deciding fit, best for complex personas. High: uses stronger reasoning models throughout, best when contacts are hard to find or the fit call is difficult. | Low (1 credit), Medium (2 credits), High (3 credits) |
| Select what contacts would you like to find | Chooses how the per-account headcount is set: a flat number per account, or a per-persona breakdown. | Contacts per account, Specify the number of contacts per persona expertise and type |
| Contacts per account | Shown for the “Contacts per account” option. How many contacts to find at each account. | A number from 1 to 100 |
| Country (optional) | Shown for the “Contacts per account” option. Restricts the search to a country, or leave unset. | Any country, or none |
| Per-persona counts | Shown for the per-persona option. For each persona expertise and type (grouped by expertise) you set a count, with an optional country per persona. Personas come from the verticals on the accounts in the run. | A count per persona (up to 50 each); optional country per persona |
| Custom instructions | Free-text guidance for the agent. | Optional text |
For the per-persona option, at least one persona must have a count above zero, and no single persona may exceed 50.
| UI label | What it means | Values |
|---|
| Skip contacts that have already been qualified | Reuses the previous persona match instead of re-qualifying the contact. | On / off |
| Follow contacts who changed jobs | When a contact has left their company, marks them as not a fit on the original account and creates a new account for them at their current company. | On / off |
| then (step picker) | Shown when “Follow contacts who changed jobs” is on: the step that the newly created account and contact continue to. | Any other step in the workflow |
| UI label | What it means | Values |
|---|
| Skip contacts that have already been researched | Reuses existing research instead of researching the contact again. | On / off |
| Auto-include all | Runs every contact research agent and auto-includes new ones. | On / off |
| Agents | When Auto-include all is off, the contact research agents to run. | Your configured research agents; at least one required |
| UI label | What it means | Values |
|---|
| Skip contacts that already have a digital twin | Uses the existing digital twin instead of generating a new one. Copilot, Roleplay, and play personalization keep working with the existing version. | On / off |
| UI label | What it means | Values |
|---|
| Skip contacts that already have plays generated | Keeps existing plays instead of generating new ones. | On / off |
| Auto-include all | Generates every play and auto-includes new ones. | On / off |
| Plays | When Auto-include all is off, the plays to generate. | Your configured plays; at least one required |
The standalone Email/Phone Finder step finds verified contact details for contacts you already have.
| UI label | What it means | Values |
|---|
| Skip contacts that already have a verified email and phone | Uses existing contact details instead of running the search again, so you don’t pay for the same email or phone twice. | On / off |
| Search for email | Turns on email finding. | On / off |
| Email verification | Verifies found emails with third-party providers. | On / off |
| Search for phone number | Turns on phone finding. | On / off |
| Phone intent scoring | Scores phone numbers for buying intent. Available only when the phone-scoring integration is connected; otherwise disabled with a link to set it up. | On / off |
| Phone scoring model | Shown when phone intent scoring is on. | Lightning Scoring, Titan Scoring |
You must enable Search for email or Search for phone number (or both) for the step to be valid.
The Condition step is the only step in the Routing group. It does no enrichment; it routes records down different paths based on rules. The builder calls it Condition.
| UI label | What it means | Values |
|---|
| Entity type | Whether this Condition routes accounts or contacts. The filters you can build depend on this choice; changing it clears any filters already set. | Account, Contact |
| Branches | The ordered set of routes leaving this step. Each branch sends matching records to a chosen next step. | Add as many as you need |
Each branch carries:
| Branch element | What it means | Values |
|---|
| Filters | Optional. With no filters the branch always runs; the panel reads “Always runs (add filters to restrict)”. Add filters and it reads “Run when filters match:”. You can build filters by hand or prefill them from a saved view of the matching entity type. | Filter group, or none |
| next / then | The step that records take when the branch runs. It is labeled “next” while the branch has no filters and “then” once you add filters (in which case records take it when the filters match). | Any other step |
| else | Shown only once a branch has filters: the step that records take when the filters do not match. | Any other step, or none |
These two controls are not agent-specific.
| Control | What it means | Values |
|---|
| then (next step) | On every agent step, the step that records continue to after this one. | Any other step, or none |
| Make this step as starting point for this workflow | Marks the step where the workflow begins. Exactly one step is the starting point; the builder flags any other step that nothing connects to as unreachable. | On / off |