Template · Agent · Official
HubSpotSalesforceFirecrawlSlackAirtable

Every lead researched before a rep opens it

New signup or inbound lead? An agent researches the company, scores the fit against your criteria, and writes the enriched record straight into your CRM.

Get this template

Opens in AGNT and takes about a minute.

The short version

Tools it needsHubSpot, Salesforce, Firecrawl, Slack, Airtable
Setup4 steps, about ten minutes
ApprovalYours, per category — nothing is sent on your behalf unless you say so
Where it runsYour own machine. Credentials stay in a local vault.

What the agent actually does

Reps open a record that is already briefed

The moment a form is submitted or a trial starts, the agent goes looking: what the company does, roughly how big it is, what stack it appears to run, recent news worth mentioning. The rep opens a record that already has the context they would otherwise have spent fifteen minutes assembling.

Score against your ICP

Fit is judged against criteria you write in plain language — not a points table bolted onto firmographic fields. "Series A or later, technical team, already using a workflow tool" is a description an agent can actually evaluate, including the parts that only show up in prose on a careers page.

CRM stays the source of truth

Findings are written back into HubSpot or Salesforce as structured fields with the reasoning attached, so the enrichment is visible where reps already work rather than in a report nobody opens.

The tools this job needs

HubSpot

Where the work arrives. The CRM your pipeline reporting depends on. The agent watches it and reads what turns up in full, rather than matching a rule against a subject line.

Salesforce

Context the agent pulls in before deciding. The enterprise system of record for revenue. It is read, not just referenced — which is what lets the decision account for it.

Firecrawl

Context the agent pulls in before deciding. Clean page extraction for research-grade scraping. It is read, not just referenced — which is what lets the decision account for it.

Slack

Where the result lands. Team chat where operations actually surface. Nothing is written here until the agent has formed a view and, where you asked for it, you have approved it.

Set it up

  1. Connect your CRM — HubSpot or Salesforce — plus whichever form or signup source creates the lead.
  2. Write your ICP as a description, including the disqualifiers, not just the ideal.
  3. Choose which fields get written back and which stay as a note for the rep to read.
  4. Watch the first batch, correct the criteria where the agent misjudged, and let it run on every new lead.

You can see exactly what it did

Nothing happens behind your back.

Every run leaves a receipt: what the agent read, which tools it called, what it decided and why, and precisely what it changed. Anything irreversible — sending, paying, publishing, deleting — waits for you to approve it. It runs on your own machine, with your own credentials, and the whole trail is yours to read afterwards. The point is not that you trust it. The point is that you never have to.

The brief it works from

This is the actual instruction set the template installs — what the agent is told to do, and what it is told never to do. Every line of it is yours to edit in AGNT after install.

Read the full brief

You research inbound leads the moment they arrive, so a rep opens a record that is already briefed.

What to find

Start with the obvious: what the company does, roughly how big it is, what it sells and to whom. Then find the signal that actually predicts fit, which is almost never firmographic. Job postings imply the problems they are hiring to solve. Public documentation reveals their stack. A changelog shows whether they are actively building. A pricing page shows who they sell to. Read those the way a good SDR would.

Score against the user's icp, in their words

They will describe their ideal customer in prose, including the disqualifiers. Evaluate against that description rather than a points table. Then write the reasoning next to the score, specifically enough to be argued with: "strong fit — engineering-led, two automation roles open, docs reference a competitor" is useful. "Score: 8" is not, and will be ignored within a week.

Write back where the rep already works

Structured fields into the CRM, plus a short note with the reasoning. Research that lives in a report nobody opens has no value. Fill empty fields; do not overwrite anything a human entered unless explicitly told you may.

When there is nothing to find

Say so. Mark the record "insufficient public information" and move on. A confidently invented profile is far worse than an empty one — a rep who opens a call having read a fabricated detail loses the deal in the first minute, and loses trust in you permanently.

Cite your sources so any claim can be checked. Speed matters here: inbound interest decays fast, and the whole point is that the research is finished before anyone has looked at the lead.

Never contact the prospect. You prepare; the rep decides and speaks.

Shown exactly as it ships. The agent inherits whichever model you already use.

Why enrichment data alone is not enough

Data vendors return firmographics: headcount, industry code, funding. That tells you the shape of the company and nothing about whether it is a fit for what you sell. The signal that matters is usually qualitative — a job posting implying the problem you solve, a docs page revealing their stack, a changelog showing they are actively building. An agent reads those the way a good SDR would, then records what it found and why it mattered.

Scoring you can argue with

A score with no reasoning is a number people quietly ignore. When the agent writes "strong fit: engineering-led, hiring two automation roles, currently on a competitor mentioned in their public docs", a rep can agree or disagree with a specific claim. That is what makes the score usable rather than decorative — and it is also what lets you correct the criteria when it is consistently wrong about a segment.

Speed changes conversion, not just tidiness

Inbound interest decays fast. An enrichment pass that completes in a minute means the first touch can be same-hour and specific; a nightly batch means it is next-day and generic. Because the agent runs on arrival rather than on a schedule, the research is finished before anybody has looked at the lead.

Common questions

Where does the research come from?

Public sources — the company site, careers pages, docs, changelogs, news — plus any enrichment API you already pay for. Sources are recorded so a rep can check a claim.

What happens when it cannot find anything?

It says so. A record marked "insufficient public information" is useful; a confidently invented profile is worse than an empty one.

Will it overwrite data our reps entered?

Only where you allow it. The usual configuration is that agents fill empty fields and add notes, while human-entered values are left alone.

Give AI a job. Get the proof.