Deal-FlowAgent

Investment research · automated

Detailed brief on any startup or fund. Sourced, not guessed.

Name a company, a founder, a VC fund or a General Partner. The agent researches it across the venture databases and the open web, then writes the page a partner reads before Monday's meeting. Every figure carries a citation; anything the evidence does not support reads Not disclosed.

Start a brief See what's on the page
  • 2–3 minStandard depth
  • 6–10 minPremium depth
  • 1 pageGuaranteed, always

How it works

1

Answer a few questions

Startup or fund. The name — or just the founder's or GP's name, and the agent resolves the company. Standard or Premium depth. Optionally your thesis, extra metrics, and any files you have.

2

The agent researches

Structured records first, then targeted web angles for market size, competitors, team and funding. Every source attempt is logged; a dead source never kills the run.

3

You get the PDF

Written, checked twice — every available metric must appear, every figure must trace to the evidence — then laid out and fitted onto exactly one page.

Start a brief

What you get

Two page types, the same discipline behind both. The page is always complete — but a private company does not publish its retention, and most funds do not publish their returns. So each list is split by what the evidence can actually support.

Startup one-pager

For an investment decision

Always — from public records

  • Bottom line — PURSUE / PASS / WATCH with the decisive evidence
  • Problem, Solution and the "why now"
  • Market size for the company's own category, from analyst reports
  • Headcount and trend, location, founding year
  • Last round — type, amount, date — and a reasoned entry path
  • Team, investors and board seats
  • Competitors, where the evidence names them, with capital raised
  • Open diligence questions and numbered sources

When the evidence has it — press, filings, or your deck

  • ARR / MRR and growth rate
  • Net revenue retention, logo churn
  • Gross margin, pricing / ACV, CAC payback, burn multiple
  • Valuation, named customers and pilots
  • SAM and SOM, when a figure is actually stated
  • Thesis alignment — scored, when you supply a thesis

VC fund memo

For an LP allocation decision

Always — from public records

  • Bottom line — COMMIT / TRACK / PASS
  • Fund history table — every vehicle whose close was announced, with its size
  • Strategy, entry stage, geographies and sectors
  • Team — founders and partners, tenure, and any departure on record
  • Portfolio count, co-investors, and exits on record
  • Peer funds of the same strategy type
  • Risks, open diligence questions and numbered sources

When the evidence has it — GP materials, filings, press

  • Net IRR, TVPI, DPI, MOIC — and the vintage quartile beside them
  • Management fee, carry, hurdle, GP commitment, minimum ticket
  • LP base and re-up rate
  • Capital committed to date, first and final close
  • Portfolio construction — positions, reserves, cheque size
  • Mandate fit — scored, when you supply a mandate

Nothing in the second list is invented to fill a gap. A metric the evidence does not support reads Not disclosed and turns into a diligence question with the name of whoever can answer it — and the brief tells you, every time, how many of the metrics it did find made it onto the page.

Where the evidence comes from

PitchBookCompany and investor profiles, deals, industry taxonomy
CrunchbaseRounds, investors, founders, leadership, funds, exits
LinkedInCompany and people records — identity, headcount, careers
Web searchPress, analyst reports, market sizing, competitors
Your filesDeck, data room, PPM, DDQ — first-party, marked as stated

Figures from aggregators and estimate sites are labelled unverified; figures that appear only in a company's or fund's own materials are marked company-stated or GP-stated. A figure that cannot be traced back to the evidence is removed before the page is delivered.