Things people ask before signing up.

An AI agent that builds and maintains a financial model for you in five minutes, then answers what-if questions like "what if we hire 2 engineers in June?" or "what if the seed slips a quarter?" against your real numbers. Two project types: Business (founders) and Life (personal money), same engine.

fingine.ai is built and operated by Cognitive Products Labs Pvt. Ltd., an Indian private limited company. Hemant Bangar is the founder. For anything legal, billing, or grievance-related, write to hemant@fingine.ai.

Generic LLMs have no persistent memory of your business and are unreliable for runway math. fingine.ai keeps a typed model of your cash, MRR, headcount, opex, and fundraise plans, and runs every what-if through a deterministic runway calculator on the server. The LLM picks tools and explains what changed; it never computes the runway number itself.

No. The AI agent picks the right tool (set_revenue, add_headcount, etc.) and the server runs a pure-function TypeScript runway calculator on the model. Same inputs always give the same number. The agent quotes the number in its reply but never invents one.

Data is stored encrypted at rest in Supabase (AES-256), transmitted over TLS, and never used for model training. Each row is scoped by Postgres row-level security so even a leaked anon key cannot leak across tenants. You can delete your account from Settings; that cascades to delete your model, transcripts, and provenance.

USD, INR, GBP, EUR, JPY, SGD, AUD, CAD, ILS, BRL, AED, CHF, SEK, NOK, DKK, HKD, NZD, MXN, ZAR, KRW. Indian users see lakh and crore grouping; Japanese users see man/oku; European users see local thousand separators. The agent's web-search trust list is biased toward your market based on your selected currency.

Free during the closed pilot. There are usage limits to keep API costs sustainable: 60 turns, roughly 200K input + 30K output tokens per day per account. Pricing arrives after the first 50 users have validated the wedge.

Every decision card shows a "show the math" toggle that breaks down cash, MRR, headcount, opex, total monthly costs, and avg burn over the next 6 months. The math is deterministic and traceable. If the number disagrees with your spreadsheet, it usually means the underlying inputs differ. Fix the input via the all-inputs panel and the number updates.

Yes. The Life project uses the same engine but with personal-finance vocabulary: take-home income instead of MRR, fixed monthly costs instead of opex, planned big spends instead of one-time costs. Decisions like "should I take a year off?" or "can I afford this car?" get the same runway treatment.

When the agent recommends a decision, it can register up to three things to come back on later: "watch MRR over the next 3 months", "if cash drops below 6 months, freeze hires", etc. Those land on your Decision Room as cards with happened / didn't-happen / skip buttons. Marking "happened" captures a structured signal you can act on; the long-term feedback loop.

Yes, sparingly, and only against a curated allowlist of about 100 trusted sources: founder essays (Paul Graham, Sam Altman, Sive.rs), top-tier VCs (YC, a16z, Sequoia, Stripe Press, Blume, Atomico), industry research (CB Insights, McKinsey, HBR), and government/central-bank authority (RBI, SEC, Federal Reserve, Bank of England, ECB, MAS, BoJ). Never Twitter, Reddit, or generic Medium. Cited inline.

Not currently. The repo is private during the pilot. May open-source non-core components (the runway calculator, evals harness) once the wedge validates.

Settings → Account → "delete my account" → confirm. Cascades to delete your model, every transcript, every provenance entry, and every decision row. Irreversible. No 30-day grace period; the data is gone the moment you confirm.