Filters

AI Models

Data

Redis Qdrant

Platforms

Streamlit Replit Vercel

Upcoming AI Hackathons
For Innovators & Creators

Discover AI Applications

Browse applications built on modern technologies. Explore PoC and MVP applications created by our community and discover innovative use cases for modern technologies.

Probably runs on Samsung S25

Hybrid AI router that intelligently routes user queries between deterministic tools, local OpenVINO models, and Fireworks AI, optimizing for accuracy, latency, and cost while minimizing unnecessary remote API calls.

Rage Quitters

AI/ML APIChatGPTCodex+26

Confidence-Cascade Routing Agent

A highly token-efficient AI routing agent that solves tasks using a free local Gemma 4B model, only escalating to a remote 31B model via Fireworks AI when strict confidence thresholds fail.

RooAnu

AMD Developer CloudAntigravityGemma+1

Hybrid token-efficient routing agent using Python.

Cost-minimizing AI routing agent, routes each task to the cheapest sufficient model. Tries to use local models for simple tasks and escalates to Fireworks AI models if needed. Agentic task classification, model tier selection, and automatic fallback.

air

AgentOpsAMD Developer Cloud

AsterRoute: Zero Token Binary LLM Router

A local DistilBERT binary router sends each query to the cheapest suitable Fireworks model, preserving answer quality while using zero routing tokens across eight task categories.

EncryptedSoul

HuggingFace Hubrest apiBERT+4

Pact — Hybrid Token-Efficient Routing Agent

Pact routes each task to the cheapest model that can handle it — heuristic triage (0 tokens) → cascade execution (cheap → medium → powerful) → heuristic quality gate (0 tokens). Only essential Fireworks tokens, zero waste.

Loom

AMD ROCmOpenAI

ZaynDev — Token-Efficient General-Purpose Agent

A general-purpose AI agent for Track 1 that passes the accuracy gate on the fewest tokens: zero-token regex routing, strongest-model-per-category, and answer-only prompts that kill the chain-of-thought verbosity dominating the token bill.

ZaynDev

Claude Coderest apiAMD ROCm+1

Codyssey Captionate: Cognitive Video Telemetry

An aesthetic cognitive captioning workspace and real-time audio telemetry platform powered by Fireworks AI and AMD Cloud computing. Featuring multi-style subtitle tuning, instant Firestore sync, and dynamic interactive Recharts sentiment tracking.

Codyseey

AI StudioAI/ML APIAMD Developer Cloud+12

ROCmify - Automated CUDA to AMD ROCm Migration

ROCmify automatically converts CUDA codebases to AMD ROCm. It fixes what it can for free with deterministic rules, only escalates to an LLM as a last resort, and won't guess at inline PTX, catching bugs like hardcoded warp sizes no compiler can see.

Maverick_soul

AMD Developer CloudAMD ROCmClaude Code+1

HybridRouter

It is a token efficient router which uses machine learning for selecting the most efficient route for saving tokens.

2Terabyte

HuggingFace SpacesAgentOpsAMD Developer Cloud+2

shrinkprompt

hrome Manifest V3 extension that shortens AI prompts on ChatGPT, Claude.ai, and Gemini — local regex fluff removal plus optional semantic compression via a containerized companion (Fireworks Gemma) or local Ollama.

Hack Haven

GemmaGemma 2Antigravity+1

CrashLens: ML Failure Doctor for GPUs and Agents

CrashLens is an AI-powered reliability platform for AMD GPU workloads and agents that combines real-time rocm-smi metrics and Gemma AI to diagnose failures in seconds, providing root cause analysis, actionable fixes, and comprehensive agent observability.

Scorpion

GemmaAgentOpsClaude Code+3

Rune - A Hosted AgentOS

Rune is a self-hosted, event-driven AgentOS that runs autonomous AI agents on minimal hardware routing every request between a free local model and paid cloud tiers by actual complexity, so you never pay cloud prices for a "hello."

Kingdom Intelligence

AgentOpsAMD Developer CloudAMD ROCm+2

NodeDash

NodeDash models your company as a live graph every role gets an AI "Chief of Staff" agent, connected by edges that carry approvals and data. When a vendor slips, agents renegotiate the whole schedule automatically no waiting for the deadline to find out.

CloseAI

AI/ML APIAMD Developer CloudAMD ROCm+6

GemmaGate

GemmaGate is a validation-gated routing agent that solves tasks locally when it can verify the answer, then escalates to Fireworks models only when needed to minimize token usage.

Area Larpers

AI/ML API

Zero-Cost Multi-Agent Token Router

A hybrid routing agent that minimizes API costs by classifying queries on the fly, leveraging fast local models for trivial tasks, and dynamically spinning up a local multi-agent debate committee for complex problems.

The Goods

Antigravity

NoCoast AI

We built an efficient routing agent to effectively answer prompts in several categories as efficiently as possible.

NoCoast

AI/ML APIGemma

Saboteur: Chaos Monkey for AI agents

Saboteur breaks your AI agent on purpose. It injects fake API errors, rate limits, bad tool data and lost context into a running agent, then scores how well it recovers. Chaos Monkey but for AI agents.

SUNE

Llama 3.1GemmaAMD ROCm+3

router-track1

baseline submission for track1baseline submission for track1

EvolutionRouting

AI/ML API

Metis

A local-first hybrid AI agent that answers tasks with a free on-device model and escalates to paid cloud models only when a calibrated confidence gate says it must — maximum accuracy at minimum token cost.

Kingdom of Science

AMD Developer CloudAMD ROCmClaude Code+5

TokenGolf

Token-efficient Track-1 agent handling all 8 categories via one Fireworks call with hidden reasoning disabled (the dominant token cost), plus an optional free local tier and a calibrated gate that escalates only when unsure.

OMEGA

AgentOps

TriForge: Hybrid Token-Efficient Router

TriForge is a hybrid AI routing agent that dynamically selects between local models and Fireworks AI using semantic intent classification, self-consistency, and Verify-Draft verification to optimize token usage, latency, and response quality.

TriForge

AMD Developer CloudAI/ML APIAI Studio+14

SafeO

SafeO is an Arabic-aware, multi-agent cybersecurity engine that blocks malicious ERP inputs in real time powered by Gemma 3 on AMD via Fireworks AI.

KrishnaCode

AI/ML APILlama 4Llama 3.1+6

NeuralGate:Intelligent Multi-Model AI Routing Eng

A self-resilient AI gateway that uses a fine-tuned SLM to classify queries and dynamically route them to the optimal local or cloud LLM — minimising cost and latency in real time.

The Last Heuristic

Llama 3.2ChromaHuggingFace Hub+2

Anirvuim: Adaptive Agent Intelligence

Anirvium unifies autonomous execution, plan-driven reasoning, policy governance, and trajectory intelligence. Sarvagun resolves customer-support workflows; SuperTuriya traces every decision, learns from outcomes, and improves what happens next.

Anirvium AI

AMD ROCmAMD Developer CloudQwen3+4

Dual Pitcher AI Agent

A hybrid AI Agent that can handle variety of natural language tasks across multiple disciplines, using either local or Fireworks AI models as efficiently as possible.

NPA

AI/ML APIAMD Developer CloudClaude Code+5

ProcureMind AI – Enterprise Procurement Agent

An AI-powered procurement assistant that analyzes Bills of Quantities (BOQs), estimates project costs, recommends vendors, optimizes procurement decisions, and generates commercial quotations through an intelligent conversational interface.

Sloth_Devs

AntigravityAnthropic ClaudeChatGPT+4

AIchemist-agent-v4

A token-efficient general-purpose AI agent that uses rule-based task routing across 8 capability categories — factual knowledge, math, sentiment, summarization, NER, code debugging, logical reasoning, and code generation — powered by Fireworks AI.

AIchemists

Qwen3

AMDA — Verify Before You Pay

A hybrid agent that verifies its own answers locally — code execution, program-checked math, constraint-solved logic — before ever spending a paid token. Most tasks finish at zero cost.

16Bit

AI/ML APIAgentOpsAMD Developer Cloud+2

Íris — One Brain, Multiple Bodies

Íris is a persistent AI Brain — one per user, auto-provisioned — that lives across your devices. Desktop, phone and watch become interchangeable "bodies" it sees, speaks and acts through, powered end-to-end by AMD-hosted Gemma.

The Lair of Gin

AMD Developer CloudAMD ROCmAntigravity+2

Task_Blowing_Agent

An optimized wind like light, containerized AI agent with intelligent local task routing and server less API integration for efficient, token-saving execution.

Neural Knights

Gemini AINVIDIAAMD Developer Cloud

30apps loaded