AI product leader. Hands-on builder.
I lead AI, data, and platform products at enterprise scale—and use agents to build practical systems myself.
Systems I have designed and built to test ideas across agents, memory, workflow automation, and decision support.
A research brain that continuously absorbs new market information, updates its working knowledge, and turns it into conviction-weighted analysis and daily decision support.
An experiment in using agents to research customer needs, build small tools, and operate parts of a software business.
See TARS in actionA conversion tool that turns visual n8n automations into maintainable Python code for engineering teams to take forward.
A three-tier architecture for retaining context across sessions and turning scattered knowledge into durable, usable agent memory.
A research-to-publishing system with quality gates, fact-checking, A/B variants, and deduplication built in.
Why Vibe Coding Dad
I have worked across the full AI product arc: enterprise platforms, zero-to-one launches, founding a company, and building agentic systems myself. That range helps me separate durable product opportunities from impressive demos.
AI, data, and platform leadership across
Founder experience
Founded ContextSmith, an AI CRM company.
Read
I share the experiments behind the projects: what worked, what broke, and what I would do differently.
Latest writing
How a customer memory layer can power meeting prep, roadmap decisions, and stronger product evidence.
Latest writing
How persistent research and pressure testing create more value than another summary bot.
Latest writing
Why I'm moving away from scheduled heartbeats toward agents that learn from their mistakes.
Work with me
In a focused two-week sprint, we pick one valuable workflow, build the smallest useful system around it, test it with real users, and decide what should happen next.
Let's talkFor problems like research brains, agent memory, internal knowledge, and decision support.
A good signalWork depends on finding context or asking the same experts.
The working pilotOne bounded domain, representative sources, reusable skills, and guardrails tested against real questions.
For problems like scheduling, document automation, inbox workflows, and automation-to-code.
A good signalTeams rely on manual workarounds because existing tools do not fit the job.
The working pilotOne focused tool for a specific role, connected to existing systems and tested in real work.
Find the costly friction and the knowledge behind it.
Put one bounded system into use with a real team.
Measure the signal and map the path forward.
Your team keeps the pilot, the results, and a clear next step.
Let's talk
If you are looking for AI product leadership or need help turning a messy workflow into a working system, tell me what you are working on.
Prefer email? will@vibecodingdad.com