Founders hand you their cap table, their contracts, their forecasts — on trust.
Dealflow inbox · DD workbench · for venture funds
Read every deck. Miss nothing. Share nothing.
The dealflow inbox and due-diligence workbench for venture funds. Every deck, data room and cap table is analysed on your own hardware. Nothing leaves.
Your own hardwareLM Studio · llama.cpp · vLLMAir-gap tierGDPR-nativeAudit log
What reaches the fund
Forwarded decksStraight into the Dealflow Inbox
Data roomsContracts and forecasts
Cap tablesThe numbers behind the deck
Your thesesEvery deal is scored against them
Your checklistsPre-seed SAFE to M&A buy-side
RANNSAKADealflow inbox · DD workbenchrannsaka.comRuns where the documents are · Nothing leaves
What the fund gets
Structured dealsEvery field with a confidence and its slide
Re-pitch deltas“Pitched eight months ago, cap up 33%”
Red-flag passesPage-level provenance, memo export
Grounded answersEvery answer cites document and page
Cap table recomputedThe math shown on every red finding
Fabric entity graphFounders, teams and companies across your deal history
Why it exists
Decks are the most confidential thing a fund touches.
Most AI tooling ships that material to a cloud model the moment you upload it.
Rannsaka runs the model where the documents are. Cloud LLM APIs are not a fallback for document content. Ever.
How it works
Three instruments, one paper trail.
Dealflow Inbox
Forwarded decks become structured deals: company, traction, raise, valuation — every field with a confidence and the slide it came from. Scored against your theses. Re-pitches surface with the delta: "pitched eight months ago, cap up 33%".
DD Workbench
Deal rooms with versioned checklists. Red-flag passes per document type — liquidation-preference stacking, vesting gaps, IP defects — with page-level provenance and a memo export at the end.
Fabric
Every deck feeds an entity graph of founders, teams, contacts and companies across your whole deal history. "This CTO was on the team you passed on last spring" surfaces by itself — from your own paper trail, never enriched from outside.
No citation, no answer
Answers that point at the page.
Grounded Q&A
Every answer cites document and page and deep-links to the clause. If the corpus doesn't cover it, Rannsaka says so.
Cap table recomputed, not believed
If the deck claims 62% and the numbers say 80%, that is a red finding with the math shown.
Audit log
Every view, download and export is recorded.
Confidentiality as architecture
Not a policy. A property of the build.
Local OpenAI-compatible endpoint
LM Studio, llama.cpp, vLLM; you choose the model.
Org isolation enforced mechanically
A CI lint rule fails the build on any unscoped query.
GDPR-native
Founder erasure tooling, retention auto-purge, Art. 6(1)(f) basis documented.
Air-gap tier
The app runs with zero outbound network.
Every private-markets vertical · Part of the Norse suite
From pre-seed SAFE to buy-side M&A.
The workbench is deal-type agnostic; versioned checklists encode the vertical — pre-seed SAFE/CLA, seed priced rounds, Series A, and M&A buy-side for corp dev and PE add-ons. Your own checklist is one YAML file away.
- Pre-seed SAFE/CLA
- Seed priced rounds
- Series A
- M&A buy-side
Leitrvik finds companies before the databases do. Kaupsignal reads buying signals. Rannsaka — Old Norse, "to search a house thoroughly" — answers the mid-funnel question: of everything that reaches the fund, what deserves attention, and what do the documents actually say?
Leitrvikleitrvik.com
Kaupsignalkaupsignal.comTalk to us about your dealflow.
Rannsaka is rolling out with a small number of funds. Tell us about your pipeline and the hardware you run, and we will set up a demo.
Reply within one business day