I come in, read your product and your pipeline, and tell you the truth about both.
Three weeks. What the product needs. What's missing from it. What the market is actually asking for. Which skills you don't have on staff. Where the process is costing you months.
The first three weeks are the same every time.
I read the product, the roadmap and the org the way an outsider can and an insider can't. Four things come back.
The gaps in your process
Where work stalls between the decision and the build, and why the same three things slip every quarter.
The answers you're not getting
The questions your team is avoiding, or answering optimistically because nobody senior asks twice.
The steps you're skipping
The ones that look optional in month one and become the reason you miss in month four.
The people you're missing
The next two hires you actually need, and the two roles you think you need and don't.
What the read is.
- Interviews
- You, your leads, and three of your customers. Ninety minutes each.
- The paper trail
- Your decision log, the last two quarters of roadmap, and the actual work queue — not the one in the tool.
- The document
- One written read: the gaps, the unasked questions, the skipped steps, the missing people, and what I'd do about each.
- The argument
- Ninety minutes, just us, so you can push back on every line of it before you act on any of it.
Interviews. You, your leads, three customers.
The paper trail. Decision log, two quarters of roadmap, the real queue.
The document, then ninety minutes to argue with it.
Everyone says AI. Almost nobody can say which of your hours it should be doing.
I've spent the last four years building with these tools rather than talking about them: a local model stack that runs on my own hardware, a memory system that carries context between models, agent pipelines that run this business. Not demos.
The work I'm actually good at is the part before the tools. Which hours a machine should take, which ones it will quietly ruin, and what has to be written down so it keeps working after the person who built it leaves.
CCAR-F certified by Anthropic, Aug 2026
Process first, then agents·Local models, not just the API
An audit of what your team does all day, then the short list of it a machine should be doing — sized in hours per week, with the ones that would go wrong named too.
I build them against the process you actually have: intake, triage, the decision log, the reporting nobody has time for. Agents where they help. A form where a form is better.
Where it earns a place in the thing you sell, and where it's a feature you'll regret shipping. I've shipped both kinds and know the difference by the invoice.
Agents where they change the work: a spec drafted from the ticket, a review pass before yours, QA cases written from the spec, docs that stay current because nobody has to remember to update them.
Context, memory and prompts written down, so the tool still works when the enthusiast who built it moves on.
Twenty-nine years of somebody having to ship on a date they didn't set.
| 1999–2007 | Microsoft | Xbox and Xbox 360 launch teams. Six AAA titles. Patent: video game instrumentation — the kill-shot replay in Halo. |
|---|---|---|
| 2010–2012 | Kabam | Director, Engineering Services. 50+ staff, $4M budget. Agile adoption across the org. |
| 2012–2015 | Livescribe | Director, Software Engineering Services. Built the release and automation teams from nothing. |
| 2015–2017 | Verizon | Video on demand. 250+ people across five departments and two business units. |
| 2017–2020 | 8 Circuit Studios | Co-founder. Concept to launch to live operations. |
| 2020–2022 | Kinoo | Principal TPM. 120+ people, $12M. CES Innovation Award 2022. Ten awards. Every launch on time. |
| 2022–now | Independent | Insight Float (founder, 130% revenue lift). Resilient Tomorrow, Distills. Building with agents before it had a name. |
Every one of those was a product somebody had to ship on a date they didn't set. That's the job.
The twenty-nine starts two years before the table does — testing flight simulators in 1997.
Who this is for.
You're the founder.
Twenty to a hundred and fifty people. Product is the business and you're still the de facto head of product. The bottleneck isn't your team's ability — it's that every hard call ends on your desk and there's nobody senior to argue with.
You're the product leader.
You have the title and the accountability, and nobody above you who has actually done it. You don't want a coach. You want a second opinion that's allowed to disagree with you, in writing, with reasons.
Not for you if
You need someone to run the team — hire a CPO, and I'll tell you who to look for. Or you need it built — hire an agency, and I'll tell you which kind. I'll say so on the call rather than take the fee.
What I don't do
I don't take ownership of your roadmap, your team or your delivery. No standups, no performance reviews, no PMO. You keep all of that. That's what keeps the read honest.
What it costs.
Read first. Then one of two paths. Never both, never hourly.
The read
- Interviews, document review, written read
- Ninety minutes to argue with it
- You own everything in it, and owe me nothing after
The fix
- People I've worked with
- Briefed from the read, run by me
Advisory
- Slack and email in between
- I'm in the room when the call is hard
- Three clients at most, ever
More work, when the read says it's worth it
Scoped from the read, priced from what's in it. The read is most of why it exists — it tells us which of these is worth doing and which is a hobby.
- A knowledge base that answers the question once.How decisions get made here, what we chose and why, the ones we'd do differently now. Built from interviews, not a wiki template nobody opens.
- AI through the development process.Specs from the ticket, a review pass before yours, QA cases written from the spec, docs that stay current on their own.
- The internal tools nobody has time for.Intake, triage, the decision log, the reporting. Built, handed over, written down.
I publish the numbers because haggling over the first invoice is a bad way to start.
What founders ask first.
Do you take ownership?
No. No standups, no reviews, no PMO. You keep all of that — it's what keeps the read honest.
Is it ever hourly?
Never. The read is $18,000 fixed. If we keep going it's $8,000 a month. One or the other, never both.
What if I disagree with the read?
The argument. Ninety minutes, just us. Push back on every line before you act on any of it.
Who actually does the work?
I do. The interviews, the paper trail and the document are mine. One read at a time.
Where does AI fit in this?
Two places. Inside the read, where I'll tell you which of your hours a machine should be taking and which it will ruin. And as work of its own — building the tools, putting it in the product, writing it down so it survives the person who built it. The process comes first either way; that's the whole method.
What about fixing what you find?
A different engagement, and one I'll take on — scoped after the read, briefed from it, run by me.
About
I got into software in 1997, testing flight simulators, and found out I'd rather build the thing than answer questions about it. Microsoft hired me two years later.
For twenty-nine years I've been the person who owns that — the date, the release, the thing everybody else treats as somebody else's problem. I don't want to own it again. I want to be the person you call before you commit.
I live in San Jose. I write at Resilient Tomorrow and I build at Distills.
If the read is wrong, you'll know in ten minutes.
Thirty minutes on a call. You describe the problem, I tell you whether I'd be any use, and if I wouldn't, I'll tell you who would.