# Lead Agents



Point them at your [lead record](/platform/leads) and they research, resolve and
qualify it while the person is still on the site.

## The Agents [#the-agents]

| Agent          | Status      | Writes                                                                                                          |
| -------------- | ----------- | --------------------------------------------------------------------------------------------------------------- |
| **Research**   | Live        | Company context from public pages: industry, headcount, funding, hiring, launches, with the source it came from |
| **Enrichment** | Live        | Person and company fields, and one merged record where there were three captures                                |
| **Signal**     | Live        | `latest_signal`: funding, role changes, site activity                                                           |
| **Scoring**    | Live        | `lead_score`, and the rank within open leads                                                                    |
| Segmentation   | Coming soon | -                                                                                                               |
| Follow-up      | Coming soon | -                                                                                                               |
| Reactivation   | Coming soon | -                                                                                                               |

They share one record, so a run reads as one pass:

```plain text
One record · 4 agents · 7 fields written

  Research    ✓  B2B SaaS · Compliance · 480 employees
  Enrichment  ✓  maria.ilustre@vanta.com · Head of Growth
  Signal      ✓  Funding · $150M
  Scoring     ✓  lead_score 92 · ranked 1st of 3,529 open leads
```

## When They Run [#when-they-run]

**On capture.** Every new lead goes through the chain as it lands, in seconds
rather than minutes.

**On change.** The account lookup keeps running after submit, and new signals
write onto the same record rather than into another tool.

<Info>
  Re-running the chain over an existing population (after an
  [ICP](/platform/leads/icp) change or a bulk import, for example) is not
  something these docs can describe accurately yet. Ask your FDE what a re-run
  covers in your workspace before you plan around it.
</Info>

## What They Are Not [#what-they-are-not]

**Not a sequencer.** They do not send email. They produce the judgement a
[workflow](/platform/forms/workflows) or your MAP acts on.

**Not a CRM sync.** Writing what they learned into HubSpot, Salesforce or Zoho is
[System Agents](/platform/system-agents/overview)' job.

**Not a scoring matrix you hand-tune.** Fit comes from the
[ICP](/platform/leads/icp) you define once, with a weight per dimension. Intent
comes from observed behaviour.

<Columns cols="3">
  <Card title="Research and Enrichment" icon="user-search" href="/platform/lead-agents/research-and-enrichment">
    What gets resolved, and how the ICP judgement is made.
  </Card>

  <Card title="Signals" icon="signal" href="/platform/lead-agents/signals">
    Funding, hiring and news, attached to the record.
  </Card>

  <Card title="Scoring" icon="gauge-high" href="/platform/lead-agents/scoring">
    How the score is built, and how to read it without over-trusting it.
  </Card>
</Columns>