Product · 2022 – Present
The AI data analyst used by 1.5M+ people.
Two senior engineers embedded with the founding team - shipping the arc from v1 to the chat-with-your-data platform used daily.
What we built
- AI / LLM
- English → SQL formula generator
- Product
- Charts, dashboards & data pipelines
- Scope
- Full-stack - architecture, AI, frontend, deploy
Engagement
The challenge
FormulaBot started as a simple English-to-Excel formula converter. The founder had a clear vision: make spreadsheets accessible to everyone. But the product needed to evolve fast — from formula generation to a full "chat with your data" platform — without losing the simplicity that made v1 stick. They needed engineers who could own the entire stack, move at founder speed, and treat the product like their own.
How we approached it
We embedded two senior engineers directly with the founding team. No handoffs, no tickets thrown over a wall. We worked in the same Slack channels, same sprint cadence, same codebase. The initial work was migrating the formula engine to use LLMs — building a reliable English-to-SQL pipeline that could handle ambiguous user queries and return accurate results. From there, we expanded into charts, dashboards, and data pipelines — each feature shipped incrementally and validated with real user data.
The technical work
The core challenge was building an AI layer that could interpret natural language queries, generate correct SQL across multiple database dialects, and present results in a format non-technical users could actually use. We built the prompt engineering pipeline, the query validation layer, and the rendering engine for charts and tables. On the infrastructure side, we handled the migration from a monolith to a service-oriented architecture — splitting out the AI inference, data processing, and user-facing API into independently deployable services. This let us scale the AI layer independently as usage grew from thousands to over a million users.
Where it is today
FormulaBot serves over 1.5 million users worldwide and generates $50K+ in monthly recurring revenue. The product has expanded from formula generation into a full data analysis suite — users upload spreadsheets, ask questions in plain English, and get charts, pivot tables, and insights back. Our engineers are still embedded with the team, now over two years into the partnership, continuing to ship new capabilities every sprint.
The people behind this project
Client
David Bressler Drove the product vision from a simple formula tool to a full chat-with-your-data platform. Defined the roadmap, prioritized features based on user feedback, and made the key calls on when to pivot the product direction.
Zeroic
Prashant Led the technical architecture and oversaw the migration from monolith to service-oriented architecture. Designed the AI inference pipeline and coordinated the engineering strategy across the full stack.
Shankar Prasad Built the core English-to-SQL engine and the prompt engineering pipeline. Owned the AI layer end-to-end — from query parsing to result rendering — and scaled it from thousands to over a million users.
Sudhanshu Sharma Shipped the frontend experience — charts, dashboards, and the data pipeline UI. Built the real-time rendering engine for query results and handled the user-facing product polish that made complex data feel simple.
and have been with FormulaBot for 2+ years. We started as a simple English-to-SQL formula generator. It's now a 'chat with your data' product serving 1.5M+ users. They built it like it was theirs.
Want to build an AI-powered product?
We embedded with FormulaBot from v1 to 1.5M users. If you need senior engineers who ship AI products at founder speed, let's talk.
Book a call