Ship daily.Stay off the rocks.
I'm a fractional CTO. I build your engineering team a digital twin, an agentic AI team that learns your codebase and ships real code. The guardrails go in first, so you get more velocity and more reliability at the same time.
If any of this sounds familiar, you're in the right place.
A codebase nobody trusts anymore
There's little to no test coverage. Every release is a small act of faith, and the people who used to know why things work the way they do have moved on. Nobody says it in the leadership meeting, but everyone has a list of files they'd rather not touch.
A vendor carrying most of the commits
An outsourced or offshore team writes most of your code. It made sense when you set it up, and maybe it still does. But the knowledge of how your product actually works lives in someone else's building, and you're starting to wonder what happens when that contract ends.
Releases that take longer every year
Shipping used to be quick and now it isn't, and nobody can point at the one thing that changed. You want to know whether the way you build software is still the right way, and you're ready to hear an honest answer either way.
They usually show up together. The fix starts in the same place every time: a safety net under the code.
What a digital twin actually is.
An agentic AI team turned loose on a codebase with no tests just gets you into trouble faster. So the safety net goes in first, and the agents go on top of it.
- Read the code before I recommend anything, and tell you honestly what I find
- Build the safety net first: test coverage and guardrails around the code that matters most
- Stand up an agentic AI team that learns your codebase and your conventions, then ships real code into it
- Give every agent a clear role, and put a human on your team in front of every merge
- Put your engineers in the steering seat, reviewing the work instead of typing all of it
- Tell you, plainly, when the answer is "not yet"
I also take on fractional CTO retainers, technical due diligence, and architecture reviews. Most of my best work has been removing complexity someone else added.
How working with me starts.
A conversation, free.
Thirty minutes, on a call. Tell me about the code, the team, and whoever's writing most of the commits. I'll ask questions and tell you what I honestly think, including if I'm not the right person for it. No deck, no pitch, no follow-up sequence.
A look at the code, scoped clearly.
If we both want to keep going, I read the codebase before I quote anything. You'll know the scope, the cost, and what you'll have at the end before we begin.
Safety net first, then speed.
Test coverage and guardrails go in before the agentic layer goes on top. How long that takes depends on where your coverage starts. There's no retainer trap and no long-term contract.
Proof, not promises.
I've built this same way in SaaS platforms, content management systems, e-commerce, publishing, and insurance software. The problem underneath, code built without a safety net, looks the same everywhere.
A team of one that grew to fifty
At CanWest I took a digital team from one person, me, to fifty. The part nobody tells you about leading people turned out to be the same discipline that makes an agentic AI team work today.
Read the full story →A platform the team had given up on
The team was convinced they'd hit a hard capacity limit and the only way out was a multi-year, multi-million-dollar rebuild. It was one configuration change. Utilization dropped 40% and Postmedia got about eighteen months of runway.
Read the full story →A national campaign sold before it was built
Fourteen days from a national campaign launch, the plan on the table was native apps that couldn't clear app store review in time. I dropped the complicated build, found the simple one, and we launched on time.
Read the full story →
I'm Ed.
I built a team like this long before I ever did it with AI. Twenty years ago I took a digital team at CanWest from one person to fifty. The part nobody tells you about leading people (defining roles clearly, setting up a real review loop, knowing when to delegate and when to step in) turns out to be the same discipline that makes an agentic AI team actually work.
Leading agents is a leadership problem before it's a technical one. Most AI consultants have never had to learn that by running a real team first. I did, and I made plenty of the mistakes along the way.
I'm based in Ontario, Canada. I've worked with clients across North America and Europe, and I'm glad to travel when the engagement calls for it.
What's your codebase really like?
Tell me. The first conversation is free, and I'll be honest about whether I'm the right person for it.
Start the conversation