Skip to content
GTM Agent Weekly

Research inbound leads in Attio only after a free ICP filter

When a promising inbound lead arrives, this agent first checks whether the company fits your ideal customer, and only then researches it and saves the answers to the record. You stop paying to research leads you would never sell to.

Paste this into ChatGPT, Claude or your agent tool. See the prompt

  • Prospecting / Outbound
  • advanced
  • RevOps
  • SDR manager
Category
Prospecting / Outbound
Difficulty
Advanced
Est. cost
Variable, token-based agent credits plus one credit per write (sourced)

Works with

Problem

Running a paid research agent on every inbound lead burns credits on companies you would never sell to. Reps also need the research saved to the record, not lost in a run log.

Setup

An Attio workflow fires when a company is added to a high-intent inbound list. A free Filter block stops anything below the ICP threshold, then a read-only Web agent answers two to four qualification questions. A Classify record block turns the research into an A, B or C fit band, a Switch branches on it, Update record saves the summary and band, and Round robin plus a Slack message assign and notify an owner for A and B leads.

Prompt

Starter prompt by The GTM Agent Weekly, written from the setup described on this page. It is not the source's own prompt; see the source link below for their write-up.

Starter prompt · 29 lines
# ROLE
You research inbound companies that already passed a basic ICP filter and grade their fit
so only good leads get routed to a rep.

# INPUTS
- {{COMPANY}}: name, domain, and any form answers.
- {{ICP}}: a short written description of our ideal customer.
- {{QUESTIONS}}: two to four specific qualification questions (e.g. "Do they sell to B2B?",
  "Do they have a sales team?").
- {{WEB_SEARCH}}: read-only web research.

# TASK
1. Answer each question in {{QUESTIONS}} from the company's own site and public sources.
   Give a short answer and a source URL for each. Write "unknown" if not found.
2. Write a one-sentence summary of what the company does and who it sells to.
3. Assign a fit band:
   A = clear match to {{ICP}} on every question.
   B = likely match, one question unknown or borderline.
   C = poor match or mostly unknown.
4. Give a one-line reason for the band.

# OUTPUT FORMAT (JSON)
{"summary": "", "answers": [{"question": "", "answer": "", "url": ""}], "band": "A|B|C", "reason": ""}

# GUARDRAILS
- Read-only. Do not contact anyone or sign up for anything.
- Do not guess headcount, revenue or customers. Unknown is fine.
- Your workflow must save this output to the company record in every branch, and route and
  notify an owner only for A and B. Run the cheap ICP filter before this research, not after.

Replace every {{PLACEHOLDER}} before running. Works with any capable LLM and runner.

Tools

Steps

  1. Create attributes on Companies for the research summary, fit band and owner.
  2. Trigger on Record added to list for the high-intent inbound list.
  3. Put a free Filter before the agent with ICP conditions such as domain present, headcount and region.
  4. Run the Web agent as a read-only research identity with two to four specific questions.
  5. Classify into A, B or C, save research in every branch, and round-robin plus Slack-notify only A and B leads.

Results

Handbook recipe; no outcome data.

What goes wrong

Gotchas

  • A Filter placed after the Web agent still pays for research on leads you meant to skip.
  • The agent saves nothing on its own; without an Update record block the research disappears after the run.
  • Research cost is token-based and varies per run, so measure it before turning on a busy trigger.

Source and credit

80x.mintlify.app ›

Attio Workflows Handbook (80x) · vendor

Source dated 2026-10-11 · last reviewed 2026-10-11. Summarized in our own words; see the source for details.

Related.

Push person-level AI research on hot leads to reps in Slack

When a prospect shows buying signals, like visiting your pricing page, this setup sends the rep a short research summary on that person in chat. Hot leads get noticed without reps spending hours researching.

Works with Common RoomSlack

  • Prospecting / Outbound
  • beginner