Foyr AI Hackathon – Build the Future of Agentic Design IntelligenceReady to build the future of AI-native software?At Foyr AI, we’re crafting a new era of intelligent tools — where AI agents don’t just assist, they collaborate, create, and take action.This isn't just about building prompts or playing with APIs. This is your chance to help build the next generation of intelligent, agent-powered software — tools that think, collaborate, and assist users in real-time.Win the challenge, land a paid internship, and if you kill it — walk into a full-time role on our core AI team.
About Foyr AI
At Foyr, we’re reimagining software from the ground up — moving away from menus and mouse clicks toward AI-native tools that blend reasoning, language, and real-time interaction.We’ve already launched Foyr Ideate, a generative AI product that creates stunning interior design concepts from simple prompts. Now, we’re building something even more ambitious — agent-first SaaS products that work like co-designers.
Imagine Telling An AI
“Show me three minimalist layouts using natural wood and hidden lighting — but leave the wall art area blank, I’ll pick that myself.”And watching it respond, build, and adjust in real time — while inviting your creative tweaks.This is the future we’re building. And we want you to be part of it.What You’ll BuildYou’ll build a traditional application with a UI that can also be controlled by an AI agent through voice, chat, or prompt-based instructions — using MCP (Multi-Channel Protocol).Your app should be usable through direct interaction (clicks, UI actions) and through intelligent agent commands.You’ll be evaluated on how well your agent understands, maps instructions to UI actions, responds intelligently, and integrates with the core app logic.The Problem StatementCreate any traditional application that can be used via a user interface or controlled by an AI agent through MCP (Model context protocol)Your task is to build a basic but functional app with a UI — and enable it to be driven by an AI agent via prompts, voice, or chat. The same action should be possible either by the user manually using the UI or by instructing the agent.This shows how classic tools can become agent-friendly, AI-native experiences.
Here Are a Few Example Applications You Can Build
- A shape drawing canvas with standard tools (rectangle, circle, line, move, delete) — but also controllable via agent instructions like “Draw three circles of radius 20 in a row”
- A to-do list manager where you can add, complete, or sort tasks via UI or say “Mark all today’s tasks as done” to the agent
- A layout grid tool where you can add/move blocks manually, or say “Add a 2-column layout with 3 text boxes and 1 image”
The focus is on showing how a traditional, standalone app can be extended to support agentic interactions — making it more accessible and intelligent without removing manual control.
Use any tech stack you’re comfortable with. Voice input/output is optional; a simple text interface with the agent is also fine.Internship Offer
- Winner: ₹TBD + Internship opportunity at Foyr AI
- Crack the challenge and score a paid internship with direct mentorship from Foyr’s AI team — ace that, and you could lock in a full-time spot.
- Work directly on building our next-gen AI design co-pilot
What We’re EvaluatingThis challenge is designed to test your real-world AI engineering skills across the following areas:Skill AreaWhat We ExpectPython + FastAPIREST endpoints or backend agent logic or agent interface APIPrompt EngineeringClean, composable prompt chains with contextual memoryLLM IntegrationUse of OpenAI, Anthropic, or similar (OpenAI Agents SDK a plus)Agent DesignCan your agent understand instructions, map them to UI actions, and adapt to user intent?UI + Agent ParitySame action possible via manual UI and agent control — seamless dual-mode interfaceVoice / Chat I/OSupport for speech input/output or natural language chatVectorDB / MongoDBOptional memory/history trackingAI Pair ProgrammingOptional — use of tools like Cursor, KiloCode for auto-complete, suggestions, etc.Submission Guidelines
Submit The Following
- Live demo (hosted on Replit, Render, Vercel, etc. — or share setup instructions)
- GitHub Repo
- 2–3 min walkthrough video (explain flow, architecture, LLM logic)
- README that explains:
- Prompt flow and logic
- How voice input/output is handled
- Tech stack and tools used
- Optional: diagram of agent reasoning or skill modules
- Known issues or improvements
Timeline
- Hackathon duration: 1 Month
- Submission deadline: 28 August 2025
How to Apply
Step 1: ApplyStep 2: Build your voice agent prototypeStep 3: Submit your solution with the demo hosted link + code + doc + demo videoTech Stack Guidelines
- Backend: Python, FastAPI, MongoDB, Weaviate/Pinecone/FAISS
- LLM Tools: OpenAI Agents SDK, LangChain/LangGraph (optional)
- Voice Tools: OpenAI Whisper, Web Speech API, ElevenLabs, Azure Speech, etc.
- UI Application: React, HTML5 Canvas, Vue, or any frontend framework to build the traditional interface
- Agent Integration: REST APIs or WebSocket endpoints to connect agent commands to app actions
- Infra (Optional): Vercel, Replit, HuggingFace Spaces, Docker
Why Join?You’ll work with a team that’s rethinking human-computer interaction from the ground up — not as tools, but as creative, collaborative agents.If you’re excited by real AI reasoning, generative UIs, and building agent-first products — we’d love to work with you.
Note:
This is a unpaid internship.Skills: ui/ux development,fastapi,prompt engineering,rest apis,python,instructions,voice/chat interaction,app,vectordb,mongodb,agent design,building,llm integration
