Proprietary data and leading analysis on AI agents.
Security whitepaper
OpenClaw security and build process
How an OpenClaw outcome agent is built, isolated, and run on infrastructure you own and govern. Network controls, secrets handling, observability, and the questions every IT team asks first. Written for procurement and security reviewers.
Infrastructure guide
Local or cloud: where should your AI agent live?
The deciding factors between running your agent on a hosted server or a local Mac Mini. Reliability, cost, control, and the holiday test we apply to every setup. Our recommendation, with the reasoning behind it.
Architecture guide
The best of both worlds
A managed assistant in the cloud, with optional hands on your own Mac. The two-piece OpenClaw setup that combines hosted reliability with local capability, without your team becoming an IT department.
Messaging guide
WhatsApp or Telegram for your AI agent?
Choosing the right messaging channel for your OpenClaw agent. Setup time, verification, dedicated numbers, ongoing cost, and the trade-offs Meta does not advertise. A practical comparison.
Have a question that needs a paper of its own?
If your team has a question we have not answered yet, we will write one.
Most papers on this page started as a single email from a procurement team or a CTO. Tell us what you need answered in writing.
Book a scoping call →