Building FinMoon AI: automate the expert process, not the answer

Over the last few years my co-founder Shashank Sheela and I kept noticing the same pattern among friends and colleagues: smart, well-paid people making important financial decisions badly. Not because they were careless, but because the process of making a good decision is manual, tedious and easy to get wrong.

That observation became FinMoon AI, a personal finance decision-making assistant for busy Indian working professionals. This post is about the thinking behind it, and the product principles I’d apply to any AI product.

Personal finance in India is a solo sport

India has on the order of a thousand SEBI-registered investment advisors for a country of 1.4 billion people. Wealth managers won’t take your call until your portfolio crosses ₹2 crore. For most professionals, managing money is entirely self-serve, and the outcomes show it: large parts of the salaried population are under-invested in equity, carry expensive debt, and are under-insured.

When we dug into why self-managing money is hard, three problems kept coming up:

1. Every product is built to confuse. Jargon, hidden clauses, terms that change every year. A home loan, credit card or insurance policy isn’t one choice; it’s hours of fine print.
2. Decisions don’t live in isolation. A home loan changes how big your emergency fund should be. An insurance gap changes how much you can safely invest. Nobody maps those connections, so people optimise each decision separately and miss the interactions.
3. Signing is the beginning, not the end. Rates move, policies get re-priced, and the product that made sense last year may not today. Monitoring is a job nobody signed up for.

What’s missing isn’t information

The tempting move in 2026 is to wrap a chat interface around a large language model and call it a financial advisor. People already do this: they ask general-purpose chatbots about tax regimes, SIPs, home loans and insurance. The answers feel right.

They’re usually incomplete. The problem isn’t prompting; it’s the foundation. A general chatbot answers the question you asked. An expert runs a process: they ask about your income stability, obligations, dependents, liquidity needs and tax position before they say anything, then compare options against criteria that matter for your situation, then tell you what to watch after you decide.

The expert’s value isn’t the answer. It’s the process that produces the answer — and that process is manual, repeatable and therefore automatable.

That is the core product insight behind FinMoon. We don’t generate a tip. We run a structured, expert-grade decision process, personalised to your situation, and produce a report: what to compare, what risks to watch, which hidden costs matter, and what to do next.

Product principles I’d reuse anywhere

Encode the process, let the model fill it in. The workflow — which inputs to collect, which criteria to evaluate, which risks to check — is designed up front, the way an expert would approach the decision. The language model does what language models are good at inside each step: reading dense documents, reasoning over the user’s specifics, and explaining trade-offs in plain language. That split is what makes the output consistent and reviewable.

Personal context is the moat. A credit card recommendation that doesn’t know about your spending pattern is a listicle. The more the system understands your situation — safely and with consent — the less generic, and the more valuable, every decision becomes.

Design for the second decision. Because decisions interact, the product has to remember prior ones. Choosing an emergency fund structure should inform the next insurance decision. That’s a data-model problem long before it’s an AI problem.

Reduce friction to first value. We made shared decision conversations usable without a login, so someone can see the full process on a real question before committing. Sign-up is for when you want it personalised.

Google for Startups India Hub

FinMoon was selected for the Google for Startups India Hub out of more than 2,000 applications. Programmes like this are more than a badge. They were strong validation that the problem is real and that an India-specific approach — shaped by Indian products, tax rules and financial behaviour — is a genuine differentiator rather than a localisation afterthought.

What building it has taught me

I’ve spent most of my career as an engineer and engineering leader inside other people’s products. Co-founding changes the questions you ask:

• “Can we build it?” becomes “Should we, and in what order?” The technical roadmap is a sequence of bets on what users will value, not a list of features.
• Architecture choices are runway choices. Model selection, hosting and caching are line items on a very short budget. Cost-per-useful-answer is a product metric.
• Trust is the product. In finance, one confident wrong answer costs more than ten good ones earn. Evaluation, guardrails and transparency about why the system recommends something are not optional polish.

We opened the beta to our first 200 users. If you make financial decisions in India and want to make them like an expert, try FinMoon. And if you’re building an AI product in a regulated, high-trust domain, I’m always happy to compare notes.