FoundersHQ

FoundersHQ is a proof based finance copilot for startups. It tracks spend, invoices, runway and funding. It explains every insight with linked invoice and transaction IDs, not guesses.

Comment

Inspiration

Inspiration

Startups move fast but their finances often live in messy bank exports, invoice PDFs and scattered spreadsheets. Founders need clear answers like "How long is our runway" and "Which invoices should we chase first" but most tools either feel manual or use AI in a way that can make up numbers. We wanted a system that is fast, simple and trustworthy. That meant building a deterministic finance engine and using AI only to explain what the data already shows.

What it does

FoundersHQ is a financial health app for startups with four modules:

  • Spending Health: Ingest transactions and commitments, then compute metrics like total outflow, net burn, run rate and spend creep. It highlights what is driving spend and links back to the exact transactions.
  • Invoice Control: Track invoices, predict late risk, rank follow ups and log touches like emails or calls. Every action links to the invoice and its history.
  • Runway Radar: Forecast cash week by week for base and pessimistic cases, show a runway timeline and explain cash dips using evidence from invoices and transactions.
  • Funding Fit Navigator: Surface funding options from a database, rank routes by fit and show a suggested timeline with clear tradeoffs.

Across the app, explanations must cite evidence IDs like txn_* and inv_*. The AI does not invent numbers. It only turns deterministic results into plain English.

How we built it

We built FoundersHQ with a full stack setup designed for reliability:

  • Frontend: Next.js App Router with TypeScript, Tailwind and shadcn/ui for fast iteration.
  • Backend: FastAPI in Python for deterministic computations and strict validation.
  • Database: Postgres for transactions, invoices, forecasts and audit trails.
  • Workers: Celery with Redis for background jobs like CSV imports and parsing workflows.

We separated the system into a "numbers layer" and an "explain layer". The numbers layer computes metrics and forecasts using reproducible rules. The explain layer sends a structured facts payload to the LLM and validates the output so it cannot add new numbers. If the model claims a cause, it must cite the evidence IDs provided.

Challenges we ran into

  • Data quality: Transaction labels are messy, invoices vary in format and fields can be missing. We added validation, defaults and clear error messages.
  • Determinism vs AI: It is tempting to let an LLM do the math. We avoided that by making all calculations backend only and then enforcing guardrails on any generated text.
  • UI trust: A finance tool needs to feel safe. We focused on evidence links, clear metrics definitions and predictable ranking for search and action queues.
  • Background processing: Imports and parsing can take time. We used jobs and statuses so the UI stays responsive.

Accomplishments that we're proud of

  • Built a deterministic finance engine that can be reproduced from stored data.
  • Added evidence linked explanations so users can click from an insight to the exact invoice or transaction.
  • Implemented a ranked action queue for invoice follow ups with touch logging.
  • Produced runway forecasts with a weekly table and a chart for base and pessimistic cases.
  • Designed the system so new data sources can be integrated later without changing the core logic.

What we learned

  • Trust is a feature. Users need to see why a number is true, not just the number.
  • Good product design is about reducing steps. One click from an alert to the right record matters.
  • Guardrails for AI are not optional in finance. Facts payloads, validation and citations make the system safer.
  • Simple deterministic rules can beat complex models when time is short and data is messy.

What's next for FoundersHQ

We want to turn FoundersHQ into a software as a service product that startups can subscribe to.

  • SaaS packaging: Add multi tenant billing accounts, usage tracking and admin controls so each startup has isolated data and settings.
  • Payment plans: Offer clear tiers such as Starter, Growth and Scale with monthly and annual pricing. Gate features by plan, for example number of bank accounts, invoices per month, forecast horizon and team members.
  • Licensing: Provide a simple license for startups to use FoundersHQ per company with optional add ons like extra seats, advanced forecasting and priority support.
  • Onboarding and retention: Improve setup with guided import flows, sample data mode and weekly email reports that summarise health, risks and next actions.
  • Reliability: Add stronger monitoring, audit logs and backups, plus rate limits and secure key handling for production.
  • Product expansion: Add persistent notifications with dedupe, improve global search and deepen runway scenarios and milestone off track detection.

Built With

Share this project:

Updates