100x Agent Extension
AI-powered agent workspace prototype — multi-mode chat, workflow automation, data explorer, app generation, and interest-based personalization for 100x.bot.
View live prototype- Year
- 2026
- Role
- Design Engineer
- Client
- 100x.bot

Context
V1 product prototype for the 100x agent extension ecosystem.
100x Agent Extension is a comprehensive React prototype for an AI-powered productivity platform — chat with intelligent mode switching, workflow automation, interactive data exploration, and natural-language app generation in one shell.
Built from the 100X-UI Figma system with semantic design tokens, deployed to Vercel with Neon-backed auth and a full API layer for chat, memory, workflows, and data sources.
Design-engineering prototype aligned to 100x product direction.
Akshay Saini
Design Engineer
40+
Design system components
167
React components
8
Core product surfaces
The challenge
Agent products need more than a chat box — users expect workflows, data tables, generated apps, memory, and quick actions in one coherent experience. The prototype had to validate mode switching, onboarding, personalization, and production-minded patterns before the full 100x platform shipped.
What I shipped
A full agent extension shell spanning chat reliability, automation, data exploration, and generative output — not isolated demos.
- AI chat — multi-mode agent with streaming LLM responses, attachments, and chat-scoped file tracking
- Workflow builder — visual workflow creation, browser, and build mode for automation
- Data explorer — TanStack Table views with AI-powered configuration, filters, grouping, and D3/Perspective visualizations
- App generation — React/JSX apps from natural language with sandbox preview and dynamic rendering
- Quick actions — categorized one-click prompts with interest-based personalization
- Onboarding — welcome, OTP verification, and interest selection curating the workspace
- Memory & history — persistent context, notes, todos, and thread management
- Design system — 40+ components, semantic tokens, and light/dark themes



Design decisions
- Mode-aware agent shell — chat, build, data, and app surfaces share navigation without context loss
- Semantic design tokens — CSS variables for theme compatibility across light and dark modes
- Miller's Law in UI — dropdowns capped at seven items to reduce cognitive load
- Figma-to-code traceability — components reference Figma node IDs for design-dev alignment
Outcome
The V1 prototype validates the 100x agent extension UX end to end — from onboarding and personalization through chat, workflows, data tables, and app generation. It ships as an open reference implementation with Neon auth, Vercel serverless APIs, and a documented design system.