Deal intelligence for private markets.

Turn deal data into a live operating model to pressure-test your thesis before committing capital.

01 — Librarian
Ask in plain English.
Conversations persist per company. Pick up where you left off.
02 — Causal chain
What has to be true.
Every answer is a chain of conditions you can question, not a paragraph.
mainthesis.app / praxis-capital-iii / meridian-esg-partners
Meridian ESG Partners·Regulated advisory · DACH
What has to be true for this thesis to compound?
Three conditions. (1) CAC stays below €180 through FY26.92% (2) The ESRS-E1 rule ships on schedule — it drives ~38% of new service revenue.71% (3)The Rheinland acquisition closes by Q2 without margin drag above 110 bps.64%
What breaks it?
If senior-auditor attrition crosses 18%58%
Ask about Meridian…
03 — Confidence
Rendered inline.
Every claim carries a score. Low confidence is where the work is.
04 — 8-tab profile
A structured company model.
Snapshot, model, market, unit econ, growth, risks, team, thesis — always sourced.
STEP 01 — INGEST
01

Connect to messy data.

PDFs, transcripts, management decks, spreadsheets — the whole data room. Every figure and claim traced back to its source.

STEP 02 — MODEL
02

Build a live operating model.

Revenue lines, unit economics, growth vectors, risks — assembled into a structured, causal model of the target.

STEP 03 — PRESSURE-TEST
03

Find where it breaks or outperforms.

Ask anything. Answers arrive as causal chains, not paragraphs — so you see which conditions carry the thesis and which ones thin.

Weeks of manual review, or a live model from day one.

A data room, an analyst reading for weeks, a memo that freezes the moment it's written. MainThesis replaces the workflow with something the whole team can interrogate.

Analyst + Excel + memoMainThesis
IntakeAnalyst reads the data room for weeks. Notes live in their head.Every document extracted into one queryable model, in hours.
The businessA narrative memo. Assumptions buried in prose.A live operating model. Revenue, unit econ, growth, risks — inspectable.
Pressure-testingIC asks “what if?” — analyst goes back for another week.Ask anything, get a causal chain in seconds.
Confidence“We think” and “we believe”. No calibration.Every claim carries a confidence. Low-confidence regions are the diligence map.
After the dealThe thesis is a PDF. Nobody opens it again.Every deal feeds back in. The fund's decision-making compounds.

Enterprise-grade security and controls

Deal data is some of the most sensitive material a fund touches. MainThesis is architected around that fact.

Baseline

The non-negotiables. Every client, from day one.

  • 01

    Your data is never used to train models.

    Not ours. Not our providers'. Ever.

  • 02

    Strict data isolation per client.

    No cross-client sharing. Every fund in its own logical perimeter.

  • 03

    Encrypted in transit and at rest.

    TLS end-to-end. Keys scoped per environment.

  • 04

    We only touch what you upload.

    No scraping, no ambient ingestion. If it isn't in the platform, we can't see it.

  • 05

    Delete everything, on request.

    One instruction. Documents, derived models, history — gone.

Trust builders

The architectural choices behind the baseline.

  • 06

    Private-by-default architecture.

    No public model training. No shared memory layer. Isolation is the default.

  • 07

    Hosted on secure cloud infrastructure.

    AWS and GCP primitives. Least-privilege IAM, private networking.

  • 08

    Access controls & authentication.

    Role-based permissions across funds, deals and documents. Every action attributable.

  • 09

    Auditable, grounded outputs.

    Every claim links back to the page it came from. No freeform hallucination surface.

Every deal you do
makes the next one sharper.

See where the thesis breaks — before you commit.

We're working closely with a small number of investment teams to shape the product. If you're losing weeks on every new opportunity, we'd like to talk.

Request a demo →