Scout
Internal sourcing and scoring engine for Czech SMEs.
Built by Point FM. Opening up access via waitlist.
- 95 %time saved on dealflow sourcing
Type a description, Scout finds the targets.
- 99 %time saved on high-level scoring
20 companies parsed, scored, ranked vs your thesis — in 20 minutes. Roughly 100× faster than an analyst.
- ~$0.50per company analyzed
Memo, scoring, valuation, counter-thesis.
- Anthropic + Examulti-model stack
The right model for each step.
Reading a Czech SME by hand takes 15–20 minutes. Screening a sector of 200 companies takes weeks of work nobody enjoys.
Scout reads the registry, scores the company against our thesis, drafts the memo. We use it daily on our own pipeline.
01 —Add a company, any way
ARES autocomplete from IČO drives auto-fill of NACE, sector, legal form, address, and VAT status. CSV / Excel import for batch onboarding. Manual entry for non-CZ companies.
02 —Or describe what you’re looking for
Don't have a target list? Type a description — “B2B SaaS for accounting firms”, “medtech distributors with EBITDA above 30 %”. Scout expands the search, discovers .cz domains, analyzes the sites, matches companies in ARES, and queues them for scoring.
No CSV. No IČO list. Just a description.
03 —Read the registry, automatically
Scout walks Sbírka listin year by year, extracts P&L and balance sheet from XML or scanned PDF. Last 3 years, automatic. Backfill on demand. Bulk-parse 20+ companies in parallel — watch live in the jobs page.
Currently CZ. Poland, Germany, Austria — through 2026.
04 —Score against your thesis
Composite 0–100 across growth, margin, efficiency, stability. Same company scores differently for a profitable-family-business thesis than for a high-growth software thesis.
Factors include market position, comparable transactions, and AI-disruption risk for software companies.
05 —Memo, valuation, counter-thesis
Investment thesis written by Claude. Sector multiples (EV/EBITDA, EV/Revenue) with niche-level overrides. Low–base–high range.
Plus a written counter-thesis — devil's advocate against your own hypothesis.
06 —DCF on demand
Full DCF model with WACC breakdown, 5-year projections, terminal value, EV bridge, and a WACC × growth sensitivity table. Every input editable.
We ran Scout on Aukro — a Czech C2C auction marketplace, founded 2016.
Scout pulled 7 years of filings, built the full financial series, and computed four sub-scores:
- Growth57
- Margin93
- Efficiency95
- Stability60
Fair value range: 613M – 1B CZK. Median 830M CZK.
The interesting part wasn’t the numbers. It was the risk Scout surfaced on its own:
“AI shopping agents could commoditize the discovery layer of marketplace platforms within 3–5 years.”
Nobody put that in the prompt. It came from the knowledge base Scout reads continuously.
Next.js. Supabase with pgvector. Claude Sonnet for extraction and analysis. Exa.ai for site discovery. OpenAI embeddings for the knowledge base. Vercel. Sources: ARES and Justice.cz.
Built with Claude Code as a coding partner.
“The first sector screen of 200 companies used to take three people three weeks. Now I have it by next morning.”— Marek Kříž
CzechCrunch · 12 May 2026
Konec excelové otročiny. Investor si v AI nakódoval pomocníka, který za něj hodnotí i hlídá firmy →
The CzechCrunch piece floated whether Scout could become available externally. We hadn’t thought it through.
Since then, enough single-GP funds, family offices, and search funds have asked that we’re opening up access — same engine, your thesis, your knowledge base. Expanding to Polish, German, and Austrian registries through 2026.
Software works, humans decide.