shiki AI

LP
Chat UI
Manage Team UI

✨ Inspiration

Existing AI chat tools felt fragmented and repetitive — typing the same prompts over and over was tedious. We envisioned a platform where teams could collaborate seamlessly with multiple AI agents, like having specialized teammates. Inspired by the flexibility of Git branching, we aimed to enable teams to explore ideas dynamically with specialized AI support.

🤖 What it does

shiki AI is an AI team collaboration platform that lets you: • Direct requests to specialized AI agents like @claude, @gpt, and @gemini using mentions • Fork conversations like Git branches to explore different thought paths • Bookmark key messages to keep important insights organized 📌 • Upload documents and images for AI analysis 📂 • See live syntax highlighting for mentions, Markdown, and code as you type ⚡️ • Assign specialized roles to each AI agent for expert-level responses 🛠️

🛠️ How we built it • Frontend: Next.js (App Router), Tailwind CSS, and shadcn/ui for a modern UI • Backend: Custom API routes for each AI provider (OpenAI, Gemini, Anthropic) using SSE for streaming responses • Auth & Data: Firebase for authentication and Firestore for storing team and chat data, with Jotai for state management • Payments: Stripe integration for subscription management • UX: Rich animations and interactivity for a delightful user experience

⚠️ Challenges we ran into • Managing synchronous responses from multiple AI agents while sharing a common chat history • Implementing performant real-time highlighting for mentions, Markdown, and code • Designing a high-functioning, intuitive UI/UX without overwhelming users • Building a seamless, secure demo mode that doesn’t require user signup

🏆 Accomplishments that we’re proud of • Creating a collaborative environment where AI agents work together to produce better outcomes • Implementing “chat forking,” letting users branch conversations just like Git • Delivering real-time mention and Markdown highlighting for enhanced usability • Building a fully interactive demo accessible without an account

📚 What we learned • Effective AI collaboration requires tightly integrated UX, APIs, and state management • Real-time features greatly enhance user experience but require careful performance tuning • Early user feedback is invaluable for refining the product

🚀 What’s next for shiki AI • Adding more specialized AI agents • Integrating with external tools like Jira and GitHub 🔗 • Developing live cursors and comments for collaborative editing • Building analytics dashboards to visualize team and AI performance 📊 • Releasing a mobile app to enable on-the-go AI team collaboration 📱