The scariest part of being a founder isn't running low on cash. It's finding out a month after you needed to act.
You're heads-down building a product — not wrangling a spreadsheet for answers, or coaxing a chatbot into doing the math for you. So the numbers drift, and by the time you look, the moment to course-correct has already passed.
Burnless is the always-on financial companion for founders. It reads your real financials and watches your burn, your runway, your revenue — all of it, all the time — and surfaces what you need to know before you have to ask. That morning, not a month later.
It's open source, AI-native, and self-hosts in one command with an embedded database — no Docker, no Postgres to run. Your numbers stay on your machine; you bring your own AI.
curl -fsSL https://burnless.ai/install | shWhat if I hire? What if I raise? What if I cut? Ask in plain English and Burnless builds the scenario on your real model — then shows you the impact on runway, burn, and cash before you commit to anything.
Work in scenario mode to explore freely: it's a non-destructive overlay on your real data, so nothing you try touches the base case. Compare scenarios side by side, then promote the one you want to baseline.
- AI companion — a chat that reads your live financials and takes actions through tools (build a scenario, adjust headcount, edit a forecast line) instead of just talking about them.
- Forecasting — dependency-graph forecast lines with multiple methods (fixed, growth rate, per-unit, percentage-of, custom formula), resolved in topological order.
- The full model — revenue (subscription / usage / one-time / services), expenses, funding rounds with cap table, and headcount — all in one place.
- Reports — P&L, cash flow, balance sheet, runway, budget-vs-actuals, key metrics, and an AI-assisted board update, exportable for investors.
- Custom metrics & dashboard — pin what matters, define your own metrics, lay the dashboard out your way.
- Automations & MCP — schedule recurring AI runs, connect external tools over MCP, and expose Burnless itself as an MCP server other agents can drive.
- Bring any AI — a provider-agnostic layer works with Anthropic, OpenAI, OpenRouter, or fully local with Ollama.
The math is not the AI's guess. A deterministic, Decimal-precision financial engine and a traceable forecast dependency graph do every calculation; the AI operates that engine through tools. So when Burnless tells you your runway, it's a number you can trace — not a number a model made up.
Install with the shell one-liner:
curl -fsSL https://burnless.ai/install | shOr with npm:
npm install -g burnless
# or run without installing:
npx burnless startThen start your instance:
burnless startThis opens Burnless on http://127.0.0.1:2876 (port 2876 = BURN) with auto-login on the loopback interface, and drops you into the onboarding wizard. One command and you're in.
Requirements: Node ≥ 20.9. Don't have it (or the right version)? Pass --with-node and the installer provisions a pinned Node for you:
curl -fsSL https://burnless.ai/install | sh -s -- --with-nodePrefer to read it first?
curl -fsSL https://burnless.ai/install | less, then run it.
Your financial model is some of the most sensitive data your company has. Burnless is built so it never has to leave your control:
- Nothing phones home. Self-hosted, your financials and your AI keys stay on your infrastructure.
- Your data, your AI, your choice. Point it at any provider — or run a local model with Ollama — and own the whole stack.
- Open forever. The AGPL license keeps improvements open: anyone who ships a modified Burnless shares those changes back. It's honest, not bait-and-switch.
- It can't be taken away. Founder finance tools have a habit of getting acquired and shut down. An open-source one you run yourself can't be — the code is yours to keep, run, and reshape.
A managed cloud edition — the same codebase with the operational pieces handled for you — is on the way for teams that would rather not run it themselves. The core stays open either way.
The self-host story is the whole point: a single artifact with an embedded database — no Docker, no Postgres to run.
curl -fsSL https://burnless.ai/install | sh # or: npm install -g burnless
burnless start- Requirements: Node ≥ 20.9 (or pass
--with-nodeto provision a pinned Node). - Your data lives under
~/.burnless/data. Back it up by copying that directory — it sits outside the versioned install, so it survives updates and rollbacks. - Updates are atomic:
burnless updateswaps in the new version and runs a health check; if it fails to come up, it automatically rolls back to the prior one. Your data is never touched. - Loopback by default: Burnless binds to
127.0.0.1— that's what makes auto-login safe, since the loopback interface is the security boundary. Exposing it more widely is an explicit, opt-in step.
| Self-host | Cloud (coming) | |
|---|---|---|
| Who runs it | You | Managed for you |
| AI | Bring your own provider / key | Hosted AI included |
| Data | On your machine (~/.burnless/data) |
Hosted |
| Ops | One command | Zero-ops |
| Cost | Free, forever | Managed plans |
Burnless is a pnpm + turbo monorepo. Data flows from the database through the financial engine and AI layer into a Next.js app:
┌──────────────────────────┐
apps/web ────────────▶│ Next.js app + API + │
(dashboard, API, │ the dashboard UI │
middleware) └────────────┬─────────────┘
│
┌───────────────────────────────┼───────────────────────────────┐
▼ ▼ ▼
packages/engine packages/ai packages/db
pure-TS financial calc provider-agnostic LLM layer Drizzle ORM
(Decimal precision) (Anthropic/OpenAI/ PGlite local /
OpenRouter/Ollama) Postgres cloud
packages/cli → the `burnless` CLI (start / update / mcp serve / …)
packages/mcp → Model Context Protocol support (consume + expose)
packages/types, packages/ui → shared TS types + React components
apps/web— the Next.js app: dashboard, API routes, and middleware.packages/engine— pure-TypeScript financial calculations with Decimal.js precision (no I/O, no DB).packages/ai— the provider-agnostic LLM layer and chat/tool loop.packages/db— Drizzle ORM schema and queries; PGlite when self-hosted, Postgres for cloud and scale.packages/cli— theburnlessCLI that installs, starts, updates, and manages your instance.packages/mcp— Model Context Protocol support: connect external MCP tools and expose Burnless as an MCP server.packages/types/packages/ui— shared TypeScript types and React components.
Start with CONTRIBUTING.md for development setup and conventions. More documentation is on the way as the project matures.
- Questions, ideas, show-and-tell: GitHub Discussions
- Bugs & feature requests: GitHub Issues
Contributions are welcome. Read CONTRIBUTING.md to get a local instance running and learn the conventions. On your first pull request, the CLA Assistant bot will ask you to sign a one-time Contributor License Agreement — a low-friction, automated step that takes a moment.
Found a vulnerability? Please report it privately — see SECURITY.md.
Burnless is split so it's open where it matters and embeddable where it helps you:
- The server and the
@burnless/*packages are AGPL-3.0 (LICENSE) — improvements stay open. - The
burnlessCLI is Apache-2.0 (packages/cli/LICENSE) — so you can embed and script it freely.
A Contributor License Agreement exists for one reason: it lets us offer a commercial / hosted edition that funds continued development. It does not take away your rights to your own contribution — you keep them.
Happy building.
Burnless is an independent open-source project and is not affiliated with or endorsed by any third-party provider or trademark referenced above.








