Serial entrepreneur and AI system designer. Active ventures within M&A, Fitness, and Real Estate. Currently open to AI consulting engagements.
| companies_founded | 3 |
| revenue_generated | $2M+ |
| team | 30+ |
| outside_capital | $0 |
Off-market deal sourcing for PE firms, independent sponsors, investment banks, and M&A advisors. Multi-channel outbound built on a proprietary database of business owners showing exit intent, run on a platform my team and I wrote: campaigns, CRM, a 30-seat dialer, inbox triage, and reporting in one system.
Wholesale fitness equipment distribution. Commercial equipment for gyms and facilities, with quoting, orders, and fulfillment running on a platform we built for it.
Circle prospecting for residential real estate agents. When a listing goes live or closes, the surrounding homeowners get a cold email sequence from the agent who did the deal. I build the fulfillment layer behind it: everything that runs after intake, from the property and owner data around a listing to the campaigns, the mailbox pool, and the sending.
An operating system for running a company with AI agents: a kanban of objectives where every card is an autonomous coding session, with a fresh-context reviewer grading the work before it can pass. Apache-2.0, self-hosted on your own box. It is the generalized version of the system that runs mine, and I build it out for operators who want it running without spending their own month on it.
I take a small number of implementation engagements a year. The one I do now is OperationKit: I build your portable knowledge base and backfill it with the context you already have, stand up a private server for your agents to run on, connect it to whichever models you prefer, and write your first agent, skill, and tool files. You own the server, the files, and the keys. Nothing in it depends on me after I hand it over.
how the OperationKit build worksI'm an entrepreneur. I started in wholesale distribution with Weight Supply, went after M&A deal sourcing with Axia, and now build the layer underneath in the open with OperationKit. The common thread: every company runs on software I write, and a fleet of AI agents operates it day to day, from sourcing and enriching data to writing campaigns, triaging inboxes, and shipping code. Most of my week goes into making those agents better at real work.
// things I think about a lot: agentic systems that do operational work instead of demos, proprietary data as an edge, why the lower middle market still trades hands over cold calls, and owning your own infrastructure. This page is a single HTML file served off a box I control, and an agent deployed it.
Before this: Williams College, Tuck Bridge at Dartmouth, growth for early-stage software companies. I turned down the investment banking route and built my own things instead.
Best email is a few sentences on what you run and what you want the agents doing.
m@mikeluka.com