A Former Accountant Is Using AI to Automate Bookkeeping

A former accountant is building software to automate much of his old profession. Ahad Ali is the founder of Tabby, a real-time bookkeeping tool that handles client paperwork and keeps profit-and-loss data current, according to TechCrunch.

Ali knows the workflow from the inside. In May 2025 he led a team of 20 people handling more than 2,000 returns per year. Tabby is trying to replace that kind of operation with code.

Tabby pulls in live account data through Plaid, a service that connects bank accounts to apps, and uses AI to show a business's financial health on a live dashboard. Paperwork enters through bank feeds and document uploads. Categorizing transactions, updating the ledger, the central record of accounts, and calculating profit and loss happen continuously. The pitch is no month-end scramble.

The company is working on two products. One is a direct app for freelancers, 1099 workers and small business owners that tracks, categorizes and organizes business expenses automatically, focused on staying tax-ready without spreadsheets or manual entry, according to Tabby. The site lists tax-ready reports with one-tap export and bank-level encryption. Its tagline is "Bookkeeping that runs itself for small businesses." The other is a version for accounting firms positioned as an alternative to QuickBooks.

Fourteen months in, Tabby had 5,500 small businesses on the platform and about $100,000 in annual recurring revenue, or subscription revenue measured yearly. That is very low average revenue per account at this stage, and points to self-serve sign-up at a volume a small team could not handle by hand.

That team is seven people: Ali, his technical co-founder, four engineers and a go-to-market specialist. The company was in the process of raising $1 million in pre-seed funding, the earliest outside round.

Next is Tabby Talk, a plain-language question layer for the product. It was described as preparing to launch, not yet live.

The broader context here will be familiar to enterprise software veterans. Bookkeeping is rules-heavy, document-heavy and repetitive. It has structured inputs in bank transactions, semi-structured inputs in receipts and invoices, and a fixed output in the general ledger and tax forms. That mix fits a system pairing retrieval and classification models, AI that finds and sorts information, with accounting logic, while Plaid and similar services handle bank connections. The dashboard was never the hard part. The hard part is accuracy at the edges, catching duplicates, sorting messy merchant data correctly, and keeping a clear audit trail when the model is unsure.

In my view, the revealing choice is selling to both sides at once. Direct sales to freelancers and micro-businesses bring volume and fast feedback on sorting quality. Selling to firms as a QuickBooks alternative brings leverage, since one firm brings many books. It also brings a harder test. Firms live with reconciliation workflows, client review, year-end close and liability for errors. Low revenue across thousands of users suggests Tabby is still proving willingness to pay, and plain-language queries only help if the underlying numbers are trusted. If Talk can answer "what changed in margin this week and why" against a ledger the accountant already believes, it earns its place. If it approximates, it does not.

Looking ahead, the optimistic case starts with trust. Continuous, machine-checked books would give the smallest businesses timely visibility into cash position and true profitability, something they have rarely had. My kids grew up checking balances on phones while I still balanced a checkbook on paper, and the gap was about tooling, not diligence. Automating the tedious layer does not remove judgment from accounting. It moves human time toward review, planning and exceptions, where professionals add the most value.