Lead research brief

Sellers spend twenty minutes before every call reading the same five places. An agent can do the reading and leave the judgement, which is the split that works everywhere in sales.

The job on one screen

Runs when	A meeting is booked, or a rep opens an account with no brief in the last 30 days.
Reads	CRM record, public site, recent news, job postings, existing notes from colleagues.
Decides	What is worth saying in one page, and what it could not verify.
Produces	A brief with sources, and an explicit list of what it could not find.
Stops when	Brief written, or nothing credible found, which is itself a useful answer.
Tools it needs	Web fetch, CRM read, note store.
Autonomy to start at	Assist. It prepares, a person decides and acts.

Optional: load a real model

Run it

The fields it pulls

Required fields are never guessed. If one is missing the run stops and a person gets the document, which is the difference between an extraction agent and a wrong database row.

Field	Required	How it is found
Company	yes	`^([A-Z][\w&.\- ]+?)(?:,\|\.\| about)`
Contact email	yes	`[\w.+-]+@[\w-]+\.[\w.]+`
Headcount	optional	`(\d{2,6})\s*(?:staff\|employees\|people)`
Renewal date	optional	`(\d{4}-\d{2}-\d{2})`
Stack signal	optional	`\b(SAP\|Salesforce\|Oracle\|Workday\|NetSuite\|Snowflake\|Databricks\|AWS\|Azure\|GCP)\b`
Stated pain	optional	`problems? with ([^.]+)`

Where this one goes wrong

Confident wrong facts

Headcount and funding are the two fields models invent most. Require a source or leave the field empty.

Stale news as current

A 2023 press release read as this quarter's news makes a rep look unprepared. Always print the date next to the claim.

Brief nobody reads

Two pages of context is worse than five specific lines. Cap the length and the rep will actually use it.

How you would know it is working

Measure	Why that one
Rep edits per brief	If they rewrite it, the brief is not the one they needed.
Prep time per call	The saving you promised. Measure it once by hand before you build.
Unverified field rate	The honesty measure. Rising blanks beats rising inventions.

Earning more rope

Assist	Draft the brief into a doc a rep reads and edits.
Approve	Post it into the CRM record automatically, still read-only.
Auto	Auto-refresh briefs on a schedule. Harmless, since nothing outbound depends on it.

The value is the sourcing, not the prose. A brief where every line can be clicked back to where it came from is trusted. One without that gets skimmed once and ignored forever.

Related: CRM hygiene · Market scan · Guardrails · all agent jobs · Agent Lab home

Free from AI School - no signup, everything runs in your browser.