AI that can plan steps and use tools to finish a task — with a person in the loop where it matters.
I build agents that do real work — calling tools, querying systems, moving tasks along — with explicit checkpoints where a human decides. I'm just as interested in which steps should never be an LLM call, and often the right answer is a plain workflow with one model step inside it.
Customer support
Speech in, reasoning and tool calls in the middle, ElevenLabs speech out. Most of the engineering is about timing — a reply that arrives a beat late stops feeling like a conversation — so the pipeline is built around latency.
Operations teams
A multi-agent system that plans a task, calls the tools it needs and pauses at set checkpoints for a person to decide. The escalation points are explicit, so the agent knows what it isn't allowed to finish on its own.
Operations & internal tooling
AI steps added to the workflows a business already runs on. n8n handles triggers, integrations and retries; the model does only the part that needs a model. Much of the value was deciding which steps should never be an LLM call.
Contact
I'm open to AI engineering, architecture and training work. Tell me what you're building and what the constraints are — that's usually enough to start.