The record
What I have run, and what I have built.
Three parts of the record, kept separate on purpose: delivery leadership at scale, client automation with an operating result, and T2C's own production AI systems.
Book a callRunning delivery at scale
Problem
a 50 engineer organization across three continents, inherited late and unpredictable.
What shipped
delivery discipline, planning, staffing, and release management across six teams.
Result
13 consecutive on time releases. Defects down from about 170 to under 60. Release dates that the business could plan against, instead of hope for.
That is the operator half, and it is where the standards come from. Twenty years on the same spine got me there: UPS operations, warehouse automation and fulfillment systems, shipping and identity software, delivery management.
Client production system
The agent that helps grow the business.
Yardigans. Subscription yard care and landscaping, run as a co-op by a group of ambitious young people with no single owner.
Yardigans already had automation, and it worked: weekly admin was down to near zero and the business held even. It was a stopgap. As the business grew, the scripts multiplied and began competing with each other. A new customer lived across text messages and spreadsheets the crew ran from memory. The rules based system hit its ceiling. It could follow instructions. It could not decide.
What shipped
We put an AI agent in the middle of the operation. The same deterministic scripts became skills the agent can call, and it reasons across them to run the business while knowing the edge of its own confidence. When it is sure, it acts. When it is not, it asks the crew through the comms channel in their portal. Customers come in through the web, phone, SMS, and conversations. Jobs get verified. Invoices go out. Getting a customer, delivering the work, and getting paid all run through one agent, with a person approving every judgment call.
Result
Yardigans is gaining customers, not just holding even. The crew spends almost no time on admin or on passing information back and forth, and it is not one owner who is freed but all of them, spending their time on the customers and the work. Get Customers, Deliver, and Get Paid are live in production. Retention is the one pillar still to come. A human stays in the loop for judgment. Yardigans does not run itself. It runs on an agent that knows when to ask.
The Yardigans agent asks for approval by text before adding a service.
"The whole business now is always on, always running. We see the updates, stay in the loop as the decision makers, and spend almost all of our time focused on our customer needs."
T2C owned production systems
These are mine. I built them, I run them, and I can stand behind every detail because there is nobody between me and the system.
An AI receptionist on our own phone.
Problem
a business phone needs to answer real callers, qualify them, and capture the lead.
What shipped
a voice AI system on T2C's line that answers real calls and routes or captures the lead.
Result
it answers T2C's real calls every day and captures the lead. Call (262) 333-3510 and put it through its paces. I put it on my own number before I would put it on yours.
A multi agent pipeline with a human approval gate.
Problem
outreach at any volume means repeated sourcing, qualification, research, and drafting. Every one of those steps is a place an unsupervised system can embarrass you in front of a customer.
What shipped
a multi agent pipeline that sources and qualifies prospects and drafts personalized outreach, with a person approving every message before it sends.
Result
it runs continuously and feeds real outreach, and nothing reaches a customer without a person seeing it first. That gate is the pattern most autonomous AI demos skip and then fail on in production.
The categories matter
The delivery record came from running an organization. Yardigans is client work. The receptionist and the prospecting pipeline are T2C owned systems. I will not blur those into one portfolio claim.
Bring the problem that needs an owner
If a workflow, prototype, or AI initiative needs to reach production, I will tell you what it needs, what it does not, and whether I should be the one to own it.
Book a callThe call is free. Bring the problem and leave with a straight answer.