Play 44
Foundry Local On-Device
On-device LLM inference for air-gapped environments with cloud escalation.
On-device LLM inference for air-gapped and data-sovereign environments — local model handles routine queries, cloud escalates for complex reasoning, with automatic fallback, sync, and fleet management via IoT Hub. Supports disconnected operation with queued sync. Ideal for manufacturing floors, field operations, and government classified environments.
Architecture Pattern
Hybrid local-cloud inference: confidence-based escalation, offline queue, fleet sync
Azure Services
DevKit (.github Agentic OS)
- agent.md — root orchestrator with builder→reviewer→tuner handoffs
- 3 agents — Foundry Local Builder (gpt-4o), Reviewer (gpt-4o-mini), Tuner (gpt-4o-mini)
- 3 skills — deploy (246 lines), evaluate (178 lines), tune (233 lines)
- 4 prompts — /deploy, /test, /review, /evaluate with agent routing
- .vscode/mcp.json — FrootAI MCP with OpenAI key + cache path inputs + envFile
TuneKit (AI Config)
- config/openai.json — gpt-4o for cloud, local model config
- config/edge.json — escalation threshold, sync interval, memory budget
- config/guardrails.json — model validation, inference safety
- evaluation/eval.py — Local accuracy >80%, Escalation rate <20%
Tuning Parameters
Machine evidence
FrootAI evidence lifecycle
This is an internal evidence maturity label, not third-party certification, accreditation, legal compliance, or a production guarantee. Missing or expired evidence demotes automatically; catalog claims cannot promote a play.
This play currently has design evidence only. A runnable scenario, endpoint evaluation, and build receipts are the next contiguous gates.
Repo Intelligence
v1A no-clone, revision-pinned map for agents and humans. Observed evidence is separated from inferred flow so the output stays useful without pretending to be a full call graph.