Financial intelligence platform
Banking tracks the money. Finanly understands the business.
Finanly connects bank, card, commerce, freight and ledger activity into one reconciled financial truth — then lets you ask it questions, close the month with a human in the loop, and prove every number back to the line it came from.
Balances, transactions, payments. Where the money moved.
Orders, payouts, invoices, shipments, fees, counterparties — from the systems the money came through.
Every deposit explained, every entry balanced, every statement in parity with the ledger.
Plain-English questions answered from governed measures, with the plan and the rows attached.
Approve, post, split, undo. Nothing irreversible without a human, nothing without an audit event.
The problem nobody has adequately solved
A bank knows a payment happened. It cannot tell you what it meant.
The same deposit, seen two ways. On the left, what a bank statement can say. On the right, what Finanly reconstructs from the systems the money passed through — and posts to the books.
A payment happened. Who, why, what it cost, what it is worth to the business — the bank cannot say.
Source
Shopify payout #3F9A · 23 orders · Sep 9–12 · fetched by the Shopify connector, matched to the bank line by amount and date.
Decomposition
- Gross sales · 23 orders+$9,870.40
- Refunds · 2 orders−$412.00
- Processing fees−$286.21
- Dispute hold · 1 chargeback−$760.00
- Net payout$8,412.19
Ledger
Five postings in one journal entry, exported with an idempotency key — a retry can never post twice.
Business context
- COGS recognized from 23 fulfilments
- Customers in 4 states · 2 returning
- One dispute open · tracked daily until resolved
Businesses run their money across banks, cards, processors, stores, freight and accounting. Each system knows its part. Finanly is where the parts become one account of the business — reconciled, governed, and verifiable.
Three workflows, shown not described
See the work, step by step.
A question, a payout, a month-end. Each one is the real sequence the product runs, with demo data in the real shapes. Hover to pause; click any step.
One platform, six layers
Not a bookkeeping app with features. A finance stack with layers.
Connectivity at the bottom, control at the top. Each layer is a shipped part of the platform; pick one to see what lives there.
Layer 5 of 6
Governed intelligence
A private model proposes plans and categorizations; the platform validates against the semantic layer and computes the figures.
- Ask Your Books
- Three-tier AI Match
- Vendor enrichment
- Subscription intelligence
ai_bi_queryauto_categorizerai_vendor_enrichmentbusiness_subscription_insightsIntelligence you can verify
The model proposes. The platform proves.
For a financial institution, a chatbot that invents numbers is disqualifying. Finanly separates language understanding from calculation, data access and execution — four hands, each with hard limits.
The model proposes
Reads the request and the governed vocabulary; returns a semantic plan or a categorization suggestion with a confidence.
Never reads raw ledger rows. Never calls a connector.
The platform validates
Checks every measure, dimension and period against the registry; binds the tenant from the session; fails closed on anything unknown.
Never trusts a field the model invented. Never widens a tenant boundary.
The engine calculates
Executes the validated plan on the canonical store and returns rows, totals and the plan that produced them.
Never computes outside governed measures. Never returns a figure without its rows.
A human approves
Accepts, edits, splits or rejects; approves export batches to the ledger. Every decision is reversible and audited.
Nothing irreversible happens without this step.
What the model can and cannot do
- It receives the question plus governed vocabulary as tools: measures, dimensions, the chart of accounts, a period resolver.
- It returns a semantic plan or a categorization suggestion with a confidence and a reason.
- It never sees raw ledger rows, never calls a connector, never posts an entry.
- It runs privately on your hardware behind an internal-only gateway; every call carries a short-lived token bound to one company.
Built for real operation
Four companies close on it every month. The numbers below are read from the system.
Finanly has run the books of four operating businesses since early 2026 — two currencies, commerce and freight reconciliation, statements in parity with the ledger. Not four enterprise customers; one platform doing real work.
Counts read from the production database and live services on 10 October 2026, rounded down. No customer names, no projections.
Bigger than one application
An intelligence layer that could sit inside the products businesses already use.
Finanly is built as a platform: an API-first core, isolated connectors, a canonical data model, a governed semantic layer and a controlled model layer — deployable on infrastructure you choose. The same architecture that runs a company's close could run behind a bank's business app.
FAQ
Straight answers
What is Finanly, in one sentence?
A financial intelligence platform that connects bank, card, commerce, freight and ledger activity into one reconciled account of a business, lets people ask it questions, and runs a month-end close a human controls.
Is it accounting software?
It includes the accounting workflows — categorization, reconciliation, invoicing, statements, exports to a ledger of record — but the product is the layer that makes those workflows and the intelligence on top of them possible. ERPNext is the ledger of record today.
How do you stop the AI from making numbers up?
The model only produces a semantic plan (which measures, dimensions and periods). core-api validates it against the governed registry — 26 cubes, 124 measures, 756 dimensions — and the engine computes the figures. Every answer carries its plan and links to the rows.
Which model runs, and where?
An open-weights language model served privately on your own hardware behind an internal-only gateway. Finanly's own finance-tuned model is in development and will replace it behind the same contract. No public AI API, no cloud fallback.
Does my data leave my infrastructure?
No. The stack — core API, AI gateway, connectors, jobs, Postgres, UI — runs on a host you control. Connectors pull data in; nothing is sent to a public AI service.
Which banks and services connect?
Mercury, Plaid (thousands of banks via Plaid Link), Wise, PayPal, Shopify, Flexport, Google Analytics 4 and ERPNext are live. QuickBooks Online, Xero, WooCommerce, Stripe and others are planned, and listed as such on the connectors page.
What happens if an export runs twice?
Nothing duplicates. Every export item carries an idempotency key that is pre-checked against the ledger; a re-run is recorded as skipped. A replay with a different payload is rejected.
Can several companies live in one account?
Yes. Each company is a tenant isolated by Postgres row-level security; users hold roles per tenant; every feature is gated by plan at the endpoint and in the UI.
Is Finanly SOC 2 certified?
Not yet. The controls are in place and documented — forced row-level security, encrypted secrets, append-only audit, internal-only AI, idempotent side effects. We say so plainly rather than imply a certification.
Is this for my company or for me?
Both. Business mode covers the close, reconciliation, invoicing, inventory and statements; Personal mode brings the same engine to a safe-to-spend planner, baskets, a subscription leak detector and debt payoff.
I run product at a bank or fintech. Where do I start?
Read For Platforms for the opportunity and the honest availability matrix, then use the partnerships form. You'll get a technical overview and a first call, not a sales funnel.
How do I try it?
Ask for early access. We walk through the deployment with you — one Docker Compose stack on your host — and connect your first sources together.
See your business understood.
Private beta for businesses. Partnership conversations for platforms. Tell us which you are.