TiDB SmartLedger

“TiDB SmartLedger unifies sales, stock & scanned bills into one DB. Ask in plain English & get instant insights with data + charts. From bills to business intelligence.”

Comment

Inspiration

We imagined a store owner drowning in data — sales spreadsheets, inventory notes, and piles of scanned bills. What if they had a Robo Assistant who could instantly answer: “Who are my best customers?” or “How much stock should I reorder next week?” That vision inspired TiDB SmartLedger.

What it does

TiDB SmartLedger acts like a Robo Accountant for businesses: Ingests all data: sales, purchases, stock, customers, and even manual bills via OCR. Stores it in TiDB, a scalable distributed SQL database. Lets anyone ask plain-English questions → AI translates to SQL → results come back as data + graphs. Provides forecasts and anomaly alerts, helping the “Robo” not just answer, but also advise.

How we built it Backend: Five FastAPI microservices — Ingest, Query, Analytics, Upload (OCR), and Webhook. Database: TiDB as the single source of truth. AI: OpenAI for NL→SQL queries, Prophet for forecasting, Tesseract for OCR. Infra: Docker + Compose for portability across Windows dev and AlmaLinux VPS. UI: React page for manual CSV/file uploads; curl for quick testing.

Challenges we ran into

Making OCR text clean and reliable from noisy scanned bills. Designing safe NL→SQL pipelines that don’t allow malicious queries. Securing multiple microservices with JWT authentication. Packaging everything so it runs smoothly on both Windows and Linux VPS.

Accomplishments that we're proud of

Built a modular Robo-style system in just days. Enabled multi-ingestion: APIs, CSV, webhooks, and OCR. Delivered natural language querying with graphs so even non-technical users get insights. Deployed successfully on a real VPS — not just a local demo.

What we learned

How to combine TiDB’s scalability with AI-driven interfaces. The importance of microservice separation for hackathon flexibility. How OCR + structured DB + AI together unlock real-world use cases. Deployment discipline: same code runs on Windows dev and VPS production.

What’s next for TiDB SmartLedger

Add real-time dashboards powered by TiDB streams. Improve OCR with cloud AI APIs for receipts and invoices. Enhance Robo’s intelligence: fraud detection, churn prediction, auto-reordering. Offer it as a ready-to-use assistant for SMEs, turning every store into a data-driven store.

✨ In one line: TiDB SmartLedger is your Robo Accountant — transforming messy bills and data into clear, actionable business insights.

Built With

Share this project:

Updates