Context
Individu lets users chat naturally and receive visual, actionable responses (cards, charts, tasks) instead of plain text walls. The product is actively developed in the private jeremybosma/individu repository with desktop and mobile wrapper projects.
The problem
Personal AI assistants fail when they only chat. Users need execution across calendar, email, payments, GitHub, and more, with trustworthy auth, billing, and UI that renders structured output safely.
How we approached it
The app uses Next.js with the AI SDK (including Amazon Bedrock and RSC patterns), json-render for generative UI components, Streamdown for rich message rendering, Better Auth with Polar for subscriptions, and Drizzle with PlanetScale migration tooling. Separate wrapper repos target desktop and mobile distribution.
What shipped
- Core SaaS application (jeremybosma/individu)
- Desktop wrapper (wrappers/desktop in repo)
- Mobile wrapper with Capacitor-style sync scripts
- Database migration and production ops scripts
- Integrations with Tavily and workflow orchestration (@workflow/ai)
Outcomes
- Natural-language task execution across connected services
- Visual responses with cards and actionable UI instead of text-only replies
- Multi-surface delivery plan: web plus desktop and mobile wrappers
- Production-grade auth, billing, and database migration discipline
Stack
- Next.js
- TypeScript
- AI SDK (Amazon Bedrock, React, RSC)
- json-render
- Better Auth
- Polar
- Drizzle ORM
- PlanetScale
- Vercel Blob & Analytics
- Tavily
- Vercel Workflow
