Loading
Case study
BizAnalyzer
Financial analytics that turns small-business accounting data into statements, ratios, forecasts, valuations, and AI-assisted explanations.
Founder and builder2022 to presentLive
BizAnalyzer is the flagship app of 2diff.io, the financial-analytics platform I founded and build through my company, Second Difference Solutions. It connects to a small business's QuickBooks Online account with read-only access and turns raw accounting data into standardized financial statements, ratios, forecasts, goals, business valuations, and plain-language commentary.
I own it end to end: discovery, architecture, backend, the AI layer, billing, and deployment. Financial calculations and generated commentary need different checks, and both need to be traceable to the underlying accounting data.
My compliance background informs the validation approach. The pipeline includes monthly-to-annual reconciliation with a defined tolerance to detect parsing or classification drift. Those checks help identify inconsistencies; they do not make every source record or generated explanation correct.
The problem
Small-business owners live inside QuickBooks but rarely get analysis out of it. Turning bookkeeping into decisions has historically meant hiring a consultant or a fractional CFO, which most owners cannot justify.
The hard part is not the dashboard, it is trusting the numbers. Every business names its accounts differently, so two companies with identical economics produce completely different report JSON. And an AI that confidently invents a dollar figure on a financial statement is worse than no AI at all.
What I built
The core is a layered financial pipeline: standardize the statements, derive the ratios, manage the periods. It normalizes inconsistent QuickBooks report JSON into one comparable model and is validated against real multi-year datasets, including a check that the monthly figures reconcile back to the annual statement.
Above that sits a multi-provider AI layer over Claude, OpenAI, and Gemini, with an anti-hallucination contract: the model receives a structured financial context and is instructed to omit what it does not have rather than estimate it. The platform runs serverless on Firebase and Google Cloud with Python Cloud Functions, encrypted OAuth tokens, Stripe billing, and a static-export Next.js front end.
Validation and controls
- Fixture tests exercise parsing and classification against multi-year company data, including reconciliation between monthly figures and annual statements.
- A monthly-to-annual reconciliation with a defined tolerance guards the pipeline: monthly figures must tie back to the annual statement. It is the same detective-control discipline from my years in broker-dealer supervision.
- The AI layer receives structured financial context and instructions to omit missing values. Generated commentary still needs review against the source statements.
- The Intuit Reports API v2 migration went live behind a one-line rollback flag, so production could revert instantly if the new API misbehaved.
- Accounting access is read-only, OAuth tokens are encrypted at rest, and Firestore rules are default-deny with server-only fields.
- 11
- AI models across 3 providers
- 20+
- financial ratios computed
- 6
- business-valuation methods
Highlights
The parts worth pointing at
Where the engineering decisions actually mattered.
- Multi-provider AI with structured financial context
- Claude, OpenAI, and Gemini integrations receive context selected for the question, with instructions to omit missing data. The financial statements remain the reference for checking an explanation.
- A standardized financial pipeline
- A layered engine that normalizes messy QuickBooks report JSON into comparable statements and ratios, fixture-validated against real 2021 to 2026 data with monthly-to-annual reconciliation checks.
- Shipped an Intuit Reports API v2 migration
- Moved every customer onto Intuit's new Reports API ahead of the old one's sunset, flipped live in production behind a one-line rollback flag.
- Security and marketplace compliance
- Fernet-encrypted OAuth tokens at rest, default-deny Firestore rules with server-only fields, host-header filtering, rate limiting, and strict Intuit trademark handling.
Built with
- Next.js (static export)
- TypeScript
- Tailwind CSS
- Firebase / Google Cloud
- Python 3.11 Cloud Functions
- Firestore
- Firebase Auth + custom claims
- QuickBooks Online OAuth
- Anthropic, OpenAI, Gemini APIs
- Stripe
- Chart.js
Matthew Dettman is a graduate of the Goldman Sachs 10,000 Small Businesses program, which trains small-business owners on the same kind of financial fundamentals BizAnalyzer is built around.
Explore the sample financials
Open the free demo, review the financial statements, and compare an AI explanation with the sample numbers. No account connection is required.