The Problem with Generic AI Assistants
Every AI tool I've used in the past year has the same blind spot:
It doesn't know what I'm building.
Ask GitHub Copilot to help debug a FastAPI error and it gives you a correct-but-generic answer. It doesn't know you're running inside a RapidKit workspace. It doesn't know which modules you've installed. It doesn't know your project uses DDD architecture.
You end up spending as much time explaining context as you save generating code.
That's what we wanted to fix.
What Is Workspai?
Workspai is the AI-powered VS Code extension built on top of RapidKit — the open-source workspace platform for backend development. Learn more at workspai.com.
Think of it as: your workspace knows everything about your backend setup, and the AI taps into that.
Four AI Features That Use Your Workspace Context
✦ AI Create
Describe a project in plain language. Workspai selects the right kit, suggests relevant modules, pre-fills configuration. Works for FastAPI, NestJS, Go.
"A FastAPI authentication service with JWT, Redis sessions, and email verification"
→ Kit: fastapi.standard
→ Suggested modules: jwt-auth, redis-cache, email-smtp
→ Project scaffolded with modules pre-selected
✦ AI Debug Actions
See a diagnostic error in your Python/TypeScript/Go file? The lightbulb now includes:
- ✦ Debug with Workspai AI — pre-fills the error + workspace context
- ✦ Explain error with AI — plain language explanation with fix suggestions
The difference: it knows you're on FastAPI 0.115 in a RapidKit workspace on Python 3.12. The fix it gives you actually applies.
✦ Doctor Fix with AI
The Workspace Health sidebar runs a system check. When it finds an issue, a ✨ button appears next to it.
Click → AI gets the full issue context → suggests exact fix steps.
No more copy-pasting error messages into ChatGPT.
✦ Module Advisor
When creating a project, describe what you're building. Workspai suggests which modules fit your stack, which are compatible, which are already installed elsewhere in your workspace.
Under the Hood
Uses GitHub Copilot's local language model API (vscode.lm). No separate API key. No data beyond what Copilot normally sends. Works with GPT-4o, Claude Sonnet, or whatever model you have access to.
The AI receives structured workspace context plus targeted file excerpts. Project types, installed modules, health status, selected errors, and key entry files. Focused, not noisy.
What Stayed Unchanged
RapidKit's workspace platform is exactly the same:
npx rapidkit create workspace my-backend
cd my-backend
npx rapidkit create project fastapi.standard auth-api
npx rapidkit create project nestjs.standard notifications
All kits. All modules. All CLI commands. Unchanged.
Workspai is a product layer built on top — specifically for VS Code users who want the AI layer.
Install
Search Workspai in VS Code Extensions, or:
Publisher: rapidkit · Extension ID: rapidkit.rapidkit-vscode
Requires GitHub Copilot for AI features.
🔗 workspai.com
🔗 VS Code Marketplace
🔗 getrapidkit.com
🔗 GitHub
