Essays, playbooks, and product thinking on multi-agent systems, context rails, and autonomous development.
We benchmarked RelayFlow against a raw coding agent on Terminal-Bench 2.0. The first result was a tie, and the reason was two bugs, both of which were quietly working against us. Here is what we found, what we fixed, and the honest number that came out the other side.
An API key isn't enough to let someone else's agent into your systems. We ran the alternative live with the Ratify Protocol team: their permission slip, our agent, and a kill switch they held from the other side of the world.
We built a demo with the Ratify Protocol team to see if an agent can prove what it's allowed to do when it shows up at someone else's system. Here's what worked, and the part we got wrong.
Claude workflows are great, but they do not let you mix agent runtimes. Here is a short playbook for Claude's recommended multi-agent strategies using Agent Relayflows, with compact code examples.
Agents are still underused as disposable sidecars. The next shift is persistent agent teams with communication, scoped trust, and evidence built into the workflow.
Most cloud agents wait to be asked. Relayfile changes that: connect your providers once and the agent wakes up automatically when something in the world changes.
Your software should coordinate. Agent Relay is an open-source SDK that gives agents real-time, cross-provider communication — so you stop being the message bus in the middle.
Cognitive science, communication theory, and software teams all point to the same conclusion: agents need what humans need — clear context, defined constraints, and real-time feedback loops.
Multi-agent orchestration is a step change in how tasks get done. After 6 weeks of building with agent swarms, here's what works, what breaks, and why the planning phase becomes everything.
Agent-to-agent communication is weird, fascinating, and potentially revolutionary. Here's how we built a system where agents work autonomously while we sleep.
