Read the statement. Score the cashflow.
Lens turns an Account Aggregator deposit statement into a transparent cashflow and underwriting read — income, obligations, surplus, risk flags, and a 0–100 score. A rule-based engine computes every figure; the model only narrates.
Why
An AA deposit statement is structured data, but reading it is still manual. Is the income regular? What are the real monthly obligations? Is there a thin buffer or a bounced mandate hiding in 200 rows? Answering that by hand is slow — and easy to get wrong.
How it works
Drop the statement
A ReBIT deposit statement — XML or JSON, by file drop, picker, or paste. Raw Forge output pastes straight in. Everything is parsed locally in your browser.
The engine analyzes
Deterministic rules detect income and its regularity, group recurring obligations, compute FOIR, surplus, savings rate and balance buffer, then raise evidence-backed risk flags — bounced mandates, OD usage, thin buffer, income gaps.
Score, narrate, export
A 0–100 cashflow score from five published, weighted components, mapped to bands A–E. Bring your own key to have it explained in plain English, then export the analysis JSON or a copy-paste underwriting note.
Features
The engine decides, the AI narrates
Every number, flag, and score comes from a deterministic rule-based engine — no model in the scoring loop. The optional narration only explains what the engine already computed; it never recomputes or invents a figure.
Auditable score, not a black box
The 0–100 score is five weighted components — income regularity, FOIR health, surplus, buffer, clean record. The weights, FOIR thresholds, and flag rules are plain JavaScript you can read and change.
Every flag carries its evidence
Bounced mandates, negative balance, thin buffer, income gaps, volatile income — each risk flag links back to the exact transactions that triggered it. Nothing is asserted without a row to back it.
Bring your own key, multi-provider
Anthropic, OpenAI, Google or Groq. Your key is held in memory for the tab only, never written to disk, and sent to exactly one place — the provider you chose.
Runs fully local & offline
No backend, no telemetry, no upload. Statement data never leaves the tab. The engine, score, and flags all work with zero API key — narration is the only thing that makes a network call.
Reads XML, JSON, and Forge output
Ingests the ReBIT deposit FI schema directly and normalizes it to one internal model — byte-compatible with raw Forge output, so it pastes straight in.
Who it's for
Built for lenders, underwriters and analysts doing first-pass cashflow checks — the read you'd otherwise assemble by hand from 200 rows.
Open & honest
Read your first statement.
Drop an AA deposit statement. Get a scored, explainable cashflow read back.
FAQ