CMD+RVL FOUNDRY

For capital markets teams that run on documents

Finally get 100% quality on the pockets of dark data you actually care about.

Open-source Airlock proves the raw document never went into the model, and extraction tournaments turn each validated win into compounding quality across future extractions.

Pool 01BDC & SEC FILINGS
Pool 02CMBS FINANCIALS
Pool 03LEASE & APPRAISAL DOCS
The problem

Your data is trapped inside your documents.

01
Dark data everywhere

The numbers you need — holdings, rents, footnotes, terms — are locked inside thousands of PDFs and filings.

02
One file at a time

Asking an LLM to read one file gives you an answer. Asking it to read five thousand gives you a bill and no way to verify the output.

03
No boundary you can trust

If the document has to go into the model, the conversation can stop before it starts. Verbal assurances are not enough for capital markets workflows.

04
Back to zero every quarter

New filings arrive. The team re-does the same work. Nothing compounds.

What Foundry does
Documents in5,320SEC 10-Q filings
Foundry
Structured rows out20,238100% on the pockets that matter, with evidence → Excel
How it works

Four steps from documents to verified data.

01
Pick the pocket

Start with the field, workflow, and document family that matter most — holdings, footnotes, rate components, lease terms.

02
Get it to 100%

Multiple extraction strategies run across the document set. The one that proves it can get to 100% quality on that pocket becomes the champion.

03
Inspect the boundary

Open-source Airlock proves the document never crosses the model boundary. The model sees bounded telemetry, not raw source files.

04
Compound the win

Each validated champion raises quality and prevents backsliding, then expands to more filings and layouts over time. Next quarter starts ahead instead of back at zero.

Why not just ask an LLM?

One real example: BDC holdings extraction.

5,320 SEC 10-Q filings, Schedule of Investments. Costs vary by document type — this is one tournament we actually ran.

Re-ask ClaudeFoundry
API cost (5,320 filings)~$13,000~$50 compute
Wall clock15–90 hours~2 hours (parallel)
Model boundaryOpaqueOpen-source Airlock proof
Quality targetBest-effort100% on the pockets that matter
ReproducibilityNo — non-deterministicYes — same parser, same output
Quality compoundingStart overEach win raises the floor
Coverage compoundingOne file at a timeOne proof expands to more filings
After setup, incremental runs cost near-zero. The champion parser is already built.
Live proof

Open-source Airlock proves the raw document never went into the model.

The document stays inside a deterministic harness. The model sees bounded telemetry, not raw source files. Airlock is open source, so you can inspect exactly what did and did not cross. Then every extracted row carries its own evidence trail.

3document pools
11live extractions
36docs in current run
20,238rows extracted
One extracted row, end to end
ARES CAPITAL CORPNAV PER SHARE$19.11
SourceSEC 10-Q / Schedule of Investments
Filed2026-02-28
AirlockPASS
Routeares_v1
Championholdings.structural.v2
BenchmarkPASS
VerifyPASS
Foundry access

Get early access.

We are starting with a small set of real document pools and expanding from there.

No spam. We reach out when your pool type is live.