Your entire deal origination in Claude. On verified data.
Connect Claude to the full Belgian company database. A longlist for a new mandate, a sector study, buy-and-build targets around a platform — in minutes, with financials and directors that actually check out.
The problem
AI is fast. But not when the data is wrong.
An AI without a source invents
Ask ChatGPT for the EBITDA of a Belgian SME and you get a made-up number. AI hallucinates when it has no verified source. For an investment file, that's useless.
Longlists cost days of manual work
Building a longlist from Trends Top, reading annual accounts PDF by PDF, looking up directors. An analyst spends days on it per thesis. That time doesn't go to the deal.
Everyone sees the same teasers
Every fund sees the same teasers from the same boutiques. Proprietary dealflow starts with searching yourself, before a target hits the market — and that takes data your competitor can't pull up as fast.
How it works
Three tasks. Each in minutes instead of days.
Acquisition candidates for a file
"Give me every West-Flemish manufacturing company with EBITDA between €1M and €3M and a director over 60." Claude reads the database and returns a scored longlist. Ready for your investment memo.
Automated sector studies
"Build a sector study of Belgian refrigeration: number of players, median margin, consolidation level, largest names." Claude aggregates the full NACE sector from verified annual accounts. A by-product of your search, not a week of work.
Buy-and-build targets around your platform
"Around this platform company, find every add-on target in the same niche, 50 km radius, revenue under €5M." Claude uses lookalikes and group structures to fill your buy-and-build pipeline in a single prompt.
One database. All your AI agents plugged into it.
Your team and your AI work on the same verified source. No portals, no copy-paste, no invented numbers.
AI-native, MCP connector
Claude and ChatGPT connected directly to the database via MCP. Ask in plain language, get data from the annual accounts.
Verified source
500,000+ companies, 10 years of financials, directors, shareholders and group structures. From CBE and NBB. No hallucinations.
Sector analysis
Every NACE sector aggregated: number of players, median margins, consolidation level. A sector study in one prompt.
Lookalikes and group structures
Lookalikes based on company websites and group structures. Map the add-on candidates around every platform.
Succession signals
Our algorithm predicts director age, so you can filter on succession opportunities before a target hits the market.
Deal sheets and export
Every search becomes a scored shortlist, a Word deal sheet on A4 or a sector report. Straight out of Claude.
Two ways to start
Originate on Claude yourself, or we deliver the deals.
For your team
Put your origination on Claude
For funds that want their analysts working on verified data through Claude. We set up the connector and train your team.
- MCP connector for Claude and ChatGPT
- 10 years of financials, directors and group links
- Sector studies and longlists in plain language
- Unlimited access to the database
Done-for-you
We deliver the dealflow
Prefer the conversations ready-made? We turn your thesis into dealflow and approach owners discreetly.
- Your investment thesis translated into targets
- Discreet approach to owners
- Deal sheet per candidate: financials, red flags, motivation
- No cure no pay
Questions investors ask
Why Claude via Signumi and not just ChatGPT?
Because it runs on a verified source. A plain AI invents an EBITDA when the figure is missing. Signumi gives Claude direct access to the filed annual accounts, so what you get back comes from CBE and NBB, not from a language model.
How reliable are the financials?
Straight from the annual accounts filed with the NBB, updated daily. EBITDA per the official CBN formula. One caveat: small companies filing abbreviated accounts don't publish revenue — EBITDA and balance sheet items are there.
Can I use the output in my files?
Yes. Through Claude, every search becomes a scored shortlist, a sector study or a Word deal sheet on A4. You also export raw data to Excel for your own model.
How much work is the setup for our team?
The MCP connector takes minutes to set up; you connect Signumi to Claude or ChatGPT and ask your questions in plain language. For teams we set it up together and run a session on your own theses.
Let Claude build your next longlist. On data that holds up.
Create a free account and connect Claude to the database. Or book a demo and Emiel shows you live how a file fills up in minutes.
Create free account