VC Deal Sourcer Agent

The problem it solves

Manually trawling Product Hunt, X, Crunchbase, and AngelList for new companies eats hours of analyst time and still misses leading-edge signals. This agent runs that search automatically every weekday morning, deduplicates against everything it has already seen, and drops fresh, thesis-matched names into a shared inbox so your deal funnel feeds itself instead of running dry.

Who it's for

• Early-stage venture capital partners and principals
• VC analysts and associates who own top-of-funnel sourcing
• Solo GPs and emerging fund managers with lean teams
• Pre-seed and seed-focused investment teams
• Angel investors and syndicate leads tracking new founders

What it does

1. 1

   Build targeted search queries

   Using your fund's sectors, stage, and geography, the agent constructs four to six recency-focused searches per run and rotates query formulations to avoid repeat results.

2. 2

   Search and extract candidates

   It runs each query across Product Hunt, X/Twitter, Crunchbase, TechCrunch, LinkedIn, and AngelList, pulling company name, founders, a one-line description, stage, funding, and source URL for 20-30 raw candidates.

3. 3

   Deduplicate against memory

   It checks every candidate against seen-companies.json by exact name and root-domain match so the same company is never surfaced twice, even across weeks of runs.

4. 4

   Apply a quick thesis pre-filter

   Fast coarse filters drop obvious misses on geography, stage ceiling, and sector mismatch, and flag any founder signals you asked it to prioritize.

5. 5

   Write fresh names to the inbox

   Companies that pass the filter are appended to a dated section in inbox.txt for the Thesis Screener to evaluate, never overwriting prior entries.

6. 6

   Update its deduplication memory

   Every candidate, including filtered-out ones, is recorded in seen-companies.json so it is never re-evaluated on a future run.

7. 7

   Log the daily run

   It prints a one-line summary of candidates found, new names after dedup, and companies added, flagging when filters surfaced nothing so you can broaden your search.

Key benefits

• Fresh on-thesis companies in your inbox every weekday without manual trawling
• Catches leading-edge signals on X 2-4 weeks before formal press coverage
• Never surfaces a duplicate thanks to persistent company memory
• Tunable daily volume, sectors, stage, and source mix without re-onboarding
• Pairs directly with the Thesis Screener for an end-to-end sourcing-to-shortlist pipeline
• Flags repeat founders and other backgrounds you want to prioritize

Sample use cases

Key integrations

• Product Hunt

  Surfaces recent B2B SaaS, developer tools, and AI product launches.

• X (Twitter)

  Leading-indicator source for founder announcements, raises, and stealth-build posts weeks before press coverage.

• Crunchbase

  Authoritative funding and stage data used to validate seed and pre-seed rounds.

• TechCrunch

  Coverage of seed and Series A funding announcements and Startup Battlefield companies.

• LinkedIn

  Founder launch and stealth posts, strong for surfacing operators-turned-founders.

• AngelList / Wellfound

  Early-stage company job posts and syndicate raises that often precede press.

Getting started

1. Import the workspace — Drag the template zip into Claude Code or use File > Import Workspace to load the VC Deal Sourcer into your project.
2. Run agent onboarding — Claude starts the onboarding interview on first launch, capturing your fund's sectors, stage, geography, source preferences, and run schedule in about five to ten minutes.
3. Give it a first task — Ask it to run one live sourcing pass, review the top five results to confirm they fit your thesis, then add the provided hook so it runs each morning.

Frequently asked questions