Why we built a local-first AI IDE

We were paying $40 a month. Two subscriptions. One for an AI chat tool that "understood code." One for an IDE plugin that sent every keystroke to a server we couldn't inspect. Both were useful. Neither ran locally. And both had the same problem: we didn't own any of it.

Cancel, and it's gone. Price goes up? You pay or you migrate. Model gets lobotomized in a quiet update? Not your call.

We got tired of it. So we built Bodega One.

What is a local-first AI IDE?

A local-first AI IDE is a development environment where the AI processing runs on your own machine, not on a third-party server. Your prompts, your code, your context: none of it leaves your hardware unless you explicitly choose a cloud model. The application works fully offline. You control the compute. You control the data.

This is different from most "AI coding tools" on the market today, which are cloud-first by design. They route everything through their infrastructure, because that's where the model lives, and because your data is part of what they're selling.

The subscription math nobody talks about

$20/month sounds fine in isolation. But AI tooling has a funny way of stacking.

• Cursor Pro: $20/month
• Claude Pro: $20/month
• GitHub Copilot: $10/month

That's $600/year for three tools that all overlap in what they do. Over three years: $1,800 for software you'll never own. Miss a payment, your license pauses. Company pivots to enterprise-only? You start over.

We're not against subscriptions as a model. We're against subscriptions for tools you depend on daily and that have no offline fallback.

What your AI tool actually knows about you

When you paste code into a cloud AI tool, that code hits a third-party server. Most terms of service allow the provider to use your data to train or fine-tune models, subject to your plan tier. Enterprise plans usually opt you out. Personal plans often don't.

More practically: you're pasting your architecture, your secrets (hopefully not, but it happens), your business logic. That stays somewhere. You don't control where.

We wanted air-gap capability as a first-class feature, not an enterprise add-on. With Bodega One and a local model like Ollama, nothing leaves your machine. Not your prompts. Not your code. Not your context.

What we actually built

Bodega One is an Electron app (Windows, macOS, Linux) that combines three things we kept wishing existed in one place:

• A full IDE: Monaco editor, syntax highlighting, multi-file editing, the works.
• AI chat with context: a persistent conversation that understands your project, not just the current file.
• An autonomous coding agent with 23 built-in tools including file ops, web search, shell execution, and a 4-layer memory system. Ask it to build something. It does.

The agent runs through a Quality Enforcement Layer (QEL), a 3-level verification system that checks its own work after every file write, not just at the end. Pattern checks, compile gates, and a full structural verifier before the result reaches you.

Bring Your Own LLM: what it actually means

BYOLLM means Bodega One doesn't bundle a model. You bring yours. We ship with 10+ provider presets:

• Local: Ollama, LM Studio, llama.cpp. Full offline, zero data exposure.
• Cloud: Anthropic (Claude), OpenAI (GPT), Groq, and others, for when you want top-tier reasoning.

Switch between them in seconds. Use Ollama at home. Point at Claude when you need the best output on a hard problem. Your API keys go directly to the provider. We're not a middleman, we don't markup usage, and we don't monitor what you send.

One-time purchase. That's it.

Bodega One is $79 for Personal (2 machines) or $109 for Pro (5 machines). You pay once. You own it. Updates are included for the version you buy.

If you join the beta (opens May 1, 2026) and complete 14 days, you get a $30 promo code before launch. That brings Personal to $49. That's the only discount we offer. No flash sales, no early bird tiers, no "limited time" pressure.

We're a small team. We built something we use every day. We'd rather charge a fair price once than chase monthly recurring revenue at the cost of our users' privacy and ownership.

Try it yourself

The beta opens May 1, 2026. If you're a developer who's tired of paying to rent tools that own your data, join the waitlist.