Open Source · GB10 · Local AI

Tools for teams running
serious local AI infrastructure

Five focused tools that cover the full lifecycle — evaluating hardware, managing the network, operating the stack, tracking quota burn, and giving agents persistent memory.

Browse projects GitHub →
Projects
Five tools, one stack

Each tool does one thing and integrates cleanly with the others. All open source, all designed for operators who know what they're doing.

⏱
Web App
LapTime
Test-drive local LLM hardware before you buy. Turns benchmark numbers into something you can feel — prompt ingest, TTFT, and decode speed across hardware configs and models.
Docs → Live site → GitHub
🎛
Service
Cruise Control
Inbound bandwidth throttle for GB10 units. Web GUI with named presets, live RX/TX display, and instant apply — no SSH, no restart, no stale rules after reboot.
Docs → GitHub
🖥
Web App
AI Hub
Self-hosted operator console. Unified chat across Claude, Gemini, and Codex — plus web terminal, VS Code, and model management for Ollama across your GB10 fleet, in one browser tab.
Docs → GitHub
📊
macOS App
Runway
Real-time quota burn tracker for Claude, Codex, Gemini, and Copilot. Menu-bar gas gauge shows current burn rate and estimated time to limit — before you find out the hard way.
Docs → GitHub
🧠
Framework
AI Knowledge Base
Persistent memory for AI agents. Runtime event log, temporal search, dream cycle compression, and session handoff — so your agents don't start from zero every conversation.
Docs → GitHub
How it fits together
The full picture

Each tool is independent, but they're designed to work together. LapTime informs the hardware decision. Cruise Control manages the network once it's deployed. AI Hub is the daily operations surface. Runway keeps quota burn visible. AIKB is the memory layer underneath all of it.

Before deployment
⏱ LapTime
Evaluate hardware
↓
Infrastructure layer
🎛 Cruise Control
Network shaping
🖥 AI Hub
Operate the stack
↓
Intelligence layer
📊 Runway
Quota awareness
🧠 AIKB
Agent memory