deep-research-item

name	deep-research-item
description	Run comprehensive research on an entity — news, funding, tech stack, buyer mapping, angle building. Writes findings to the webset store and optionally to Airtable. Use after verification passes or when a channel event requires deep research on a specific entity.
argument-hint	["entityName domain websetId"]
user-invocable	true
allowed-tools	Read, mcp__websets-codemode-local__execute, mcp__websets-codemode-local__search, mcp__airtable__execute

Deep research on $0 (domain: $1, webset: $2).

Procedure

1. Parallel Exa searches

Run these searches concurrently:

• exa.search("$0 funding investors Series") — funding history
• exa.search("$0 tech stack engineering blog") — technology signals
• exa.search("$0 hiring jobs engineering") — hiring signals and growth
• exa.search("$0 news announcement 2025 2026") — recent news
• exa.findSimilar("https://$1") — similar companies (competitive landscape)

2. Content extraction

For the top 3-5 most relevant URLs from step 1, run exa.getContents to extract full text. Focus on:

• Evidence of the signal that scored this entity
• Decision-maker names and roles
• Technology choices and infrastructure
• Recent milestones or pain points

3. Buyer mapping

From enrichments and research, identify:

• Primary contact: Name, title, email (from enrichments)
• Role fit: Why this person would care about what we offer
• Org context: Team size, reporting structure if visible

4. Angle building

Based on all signals, synthesize:

• Hook: The specific signal or event that makes this timely
• Value prop: What we can offer that maps to their situation
• Risk: What could make this a bad fit

5. Store results

Annotate the webset item:

await callOperation('store.annotate', {
  itemId: '<item_id>',
  type: 'research_finding',
  value: JSON.stringify(researchBrief)
});

6. Airtable write (if target specified)

If the workflow config includes an Airtable target, upsert the research brief using the Airtable Code Mode execute tool.

7. Return brief

{
  "entity": "$0",
  "domain": "$1",
  "signals": { "funding": "...", "tech": "...", "hiring": "...", "news": "..." },
  "buyer": { "name": "...", "title": "...", "email": "...", "fit": "..." },
  "angle": { "hook": "...", "value_prop": "...", "risk": "..." },
  "confidence": 0.0-1.0,
  "sources": ["url1", "url2", "..."],
  "similar_companies": ["...", "..."]
}

Report the brief to the user.