The era of renting intelligence is over. Why pay per token? Why wait for latency? Why hand your user's data to a massive corporation?
We are entering the Local AI Revolution.
We are challenging YOU to ideate the mobile app of the future, one that runs powerful Small Language Models (SLMs) like DeepSeek R1, Llama 3, and Whisper directly on the device.
No Internet. No Fees. No Spying.
💡 The Challenge (Ideathon)This is a pure Ideathon. You do not need to submit a compiled app. We want your vision.
Using the architectural blueprints of the RunAnywhere SDK (the YC-backed standard for local AI), show us a mobile experience that is impossible on the cloud.
We are looking for ideas in 3 core areas:
-
🔒 True Privacy: Apps where user data is too sensitive to ever leave the device (Health, Finance, Journals).
-
📵 The "Offline" Edge: Tools for remote areas, subways, or disaster zones where the internet is dead.
-
⚡ Zero Latency: Voice assistants and agents that react instantly because they live on the chip, not the server.
-
🚀 Backed by Y-Combinator Founders: This event is supported by the team at RunAnywhere (YC W26). Top submissions don't just win cash; you get direct feedback and "Office Hours" with founders who are building the future of edge computing.
-
🎫 The Golden Ticket: The most visionary ideas will secure a Direct Entry slot into our upcoming Global On-Device Hackathon, bypassing the massive waiting list.
-
💰 $150 Prize Pool: Cash awards for the top 3 innovative concepts.
-
🧠 Master the New Stack: Learn how to architect apps using DeepSeek, Whisper, and Llama on a standard mobile device.
This event runs entirely on our Discord. You must join to get architectural guidance, and receive critical updates.
Requirements
We are not looking for code snippets or marketing fluff. We want to see how you think and architect a solution for the constraints of mobile hardware.
You must submit a single PDF Presentation (Slide Deck) covering the following 4 sections:
1. The Abstract (The "What" & "Why")
-
Problem Statement: What specific problem are you solving that requires on-device AI?
-
The Vision: Describe the user experience. How does the app feel instant, private, or resilient?
2. The Architecture Diagram (The "How")
-
Visual Flow: Provide a diagram showing how data moves through your app.
-
Component Usage: Explicitly state how you would use the RunAnywhere SDK.
-
Example: User Speech -> Local Whisper (STT) -> RunAnywhere Core -> Quantized DeepSeek R1 -> Local TTS.
3. The SLM Strategy (The "Brains")
-
Model Selection: Which Small Language Model (SLM) are you choosing? (e.g., Llama-3-3B for speed or DeepSeek-R1-Distill for reasoning).
-
Feasibility: Briefly explain why this model fits on a phone (mention quantization or memory usage).
4. The "Offline" Scenario
-
Describe a specific use-case where your app saves the day because it is offline. (e.g., "A paramedic using the app in a rural dead zone to diagnose a patient").
-
Project Name: Make it catchy (e.g., "GhostBot" not "Privacy App").
- Short Project Description
-
The File: Upload your PDF Presentation (Max 10 slides).
-
Discord Handle: Include your Discord username this is important we may contact you during the judging phase you must be part of the Discord Server
Prizes
Winner
$75 + Office Hours with YC Founders + Golden Ticket to Global Hackathon.
Runner-Up
$50 + Office Hours with YC Founders
Second Runner-Up
Devpost Achievements
Submitting to this hackathon could earn you:
Judges
Aditya
Judging Criteria
-
Technical Feasibility (40%)
-
Privacy & "Edge" Utility (30%)
-
Innovation & Vision (30%)
Questions? Email the hackathon manager
Invite others to compete
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
