Deploying agents inside organizations
Most of my time goes into figuring out how to deploy AI agents within real teams. Not demos, not proofs of concept. Actual agents that people use daily to get their work done.
I'm using OpenClaw, an open-source agent framework, as the foundation. Each team gets their own agent with its own identity, tools, and knowledge. A marketing team's agent is completely different from a customer support agent or a QA agent. They live on a server, they're always on, and team members talk to them through Discord or Slack like they would talk to a colleague.
The interesting part isn't the technology. It's figuring out what each team actually needs versus what sounds cool in a demo. Most agent projects fail because nobody understands the work the agent is supposed to do. I do, because I've done that work for years. That's the edge.