Blog
Thoughts on AI cost, context efficiency, and agentic development.
Governing 37 AI Agents in Production: A Field Guide to Runtime Governance.
Claude Code, Cursor, Copilot, custom agents, MCP servers — each with its own dashboard and no shared audit trail. Policy as a runtime primitive, not a doc. Six use cases from what real teams are doing.
Two things, one product: Guard and the Security Loop.
We've been telling too many stories at once. Conduct is two things: Guard on the wire in front of every AI call, and the Security Loop that closes scan-to-fix. Everything else is a playbook in the registry.
Threat modeling is a playbook, not a product.
The security industry has spent twenty years selling threat modeling as a workshop, a tool category, a compliance checkbox. It's none of those things. It's a task the AI in your PR flow can do on every merge.
Fix this code: the framing gap behind an 18-day model shutdown.
Fable 5 returned after an 18-day shutdown. The lasting lesson is how ordinary coding prompts can cross a safeguard boundary—and why enterprises need model and intent governance.
Conduct is open today. Here's the honest version.
A quiet launch note from a product-first team. What v1 is, what it isn't, how to onboard in an afternoon, and a standing offer of a 60-day free evaluation.
Which AI model should you use, and for which task?
Claude, GPT-4o, Gemini, DeepSeek, Mistral, Llama, a practical guide for developers on picking the right model for the job. With code examples and ConductGuard governance.
Stop paying Opus prices for Haiku work.
Most AI coding setups default to the most capable model for every task, including fixing a typo. That's a 4x cost penalty on work a cheaper model handles just as well.
RTK: how we cut 93% of CLI token noise from AI coding sessions.
Every time Claude runs a command, the full output lands in the context window. A single git log can burn 4,000 tokens. RTK filters that noise before it ever reaches the model.
Why your AI reads the whole file when it only needs three functions.
AI coding tools aren't lazy, they're blind. They can't see inside a file without reading it. Agent Booster gives them a map so they don't have to.