How we
work together.
I've found that agentic systems can plateau at the demo. I aim to design systems that get more useful with age: memory-first architecture, context engineering, embedded delivery, and operational discipline for regulated environments.
I've built my practice around webdevOS and cofounderOS, my in-house multi-agent orchestration engines. The four practices below are how I deliver it.
Your Claude, Dialled In
I can't get Claude Cowork to work for me.
You have Claude Cowork (and maybe Claude Code) but can't get it to work for you. It forgets context, it doesn't sound like you, and the connections to your tools feel brittle.
I dial it in across three fronts. First, memory: I make sure Claude can remember properly, so the context you build up doesn't evaporate between sessions. Second, voice: voice guidelines are paramount, so what it produces reads like you wrote it, not like a generic assistant. Third, connectivity: not all MCPs are born equal, so I wire in efficient, clean connections to the tools you actually use.
The outcome is a Claude coworker that remembers, sounds like you, and connects cleanly, so it earns a place in your day instead of being abandoned after the first week.
- D.01Memory implementation so Claude remembers across sessions
- D.02Voice guidelines so output sounds like you
- D.03Efficient MCP connectivity to your real tools
- D.04Hands-on setup and handover so you can maintain it
I have Claude Cowork but I can't get it to work for me. Can you fix that?
Yes, that's the whole point of this offering. "It doesn't work for me" almost always comes down to three things: it forgets context, it doesn't sound like you, and its connections to your tools are brittle. I dial in all three so it earns a place in your day instead of being abandoned after a week.
Why does Claude keep forgetting what I told it?
Because memory has to be implemented deliberately, it isn't automatic. I set up memory so the context you build up persists properly across sessions, which is the single biggest reason a Claude coworker starts to feel genuinely useful rather than amnesiac.
Can you make it sound like me rather than a generic assistant?
Yes. Voice guidelines are paramount in this work. I capture how you actually write and encode it so what Claude produces reads like you wrote it, which is what makes the output usable without a heavy editing pass every time.
I've connected some MCPs but it still feels clunky. What's wrong?
Not all MCPs are born equal. A clumsy or redundant set of connections makes everything feel slow and brittle. I wire in efficient, clean MCP connectivity to the tools you actually use, so the integrations help rather than get in the way.
Does this work for Claude Code as well as Claude Cowork?
It applies to both. The memory, voice, and connectivity work carries across the Claude tools you use day to day, Claude Cowork and Claude Code alike. And it isn't limited to Anthropic: I can work with non-Anthropic architectures too, so you're not locked to a single vendor.