Mike Kirkup on Building AI-Native Products at Breakneck Speed
The third Founder Dinner for W26: 1,000+ companies coached + zero lines of human-written code + four months from concept to alpha = why AI changes execution speed but not the fundamentals of building
TL;DR: Mike Kirkup, co-founder of Elderella and long-time startup coach at Communitech and once lead Velocity, shared the third dinner of our W26 series. His insight: despite coaching over 1,000 companies through the startup journey, he’s never seen product development move this fast. Elderella went from concept to alpha in four months without writing a single line of code by hand - all Claude-powered development. But the speed of building hasn’t solved the hardest problems: changing customer behavior, avoiding co-founder conflicts, and finding sustainable go-to-market motion. Mike’s building an AI-native caregiving platform while proving that experience matters more than ever when you can move this quickly.
The thrid dinner in the W26 series, thanks to the support of Osler and TD Innovation Banking, brought us Mike Kirkup, who is someone most founders in the Waterloo ecosystem know from his coaching work at Communitech and leading Velocity. Over the years, Mike has seen more than 1,000 companies navigate the early stages of building - from zero to something, zero to zero, and everything in between. But now he’s back in the founder seat himself, co-founding Elderella with Jacqui Murphy (also part of Builders Club).
What makes Mike’s current journey remarkable isn’t just that he’s building again - it’s the speed and methodology. Four months from concept to alpha. Zero lines of code written by humans. An entire mobile app, backend infrastructure, HIPAA compliance, multiplayer coordination features, WhatsApp integration, and a suite of internal tools - all built through AI-assisted development.
The Problem: Coordination in Crisis
Elderella tackles a problem most of us will face: coordinating complex situations across multiple family members when someone needs care. As Mike explained, “We know everything about our children. We saw when they grew up, we saw that they got hurt, we remember these things. But no one has any history of what their actual parents went through.”
When a health crisis hits, families scramble. Critical information lives scattered across different systems, conversations, and people’s memories. The challenge isn’t just collecting information - it’s making it actionable when it matters most.
The vision is a platform that brings fragmented information together and makes coordination across family members and care providers actually work.
The ultimate goal, as Mike put it: “Remove the statement ‘I’ll just do it because it’s easier if I do it myself.’ Expand the network of people you can lean on and make it really easy to do that.”
Four Months: Concept to Alpha
The timeline is striking:
Early September: Launched with a simple website
September-October: 75 user interviews, no product pitching, just understanding the problem
October-December: Built and launched initial product to early users
January: Expanded capabilities based on feedback
February: Preparing for broader release
This trajectory - two months from starting to build to having product in users’ hands - would have been unthinkable just a year ago. Mike contextualized it: “This same product, something similar to this, we found had been built 10 years ago. We contacted the founder. Our guess was they had 8 to 10 engineers working on it to build the same thing, and it took them probably a couple of years.”
The difference? All AI-assisted development.
The AI-Native Development Approach
The centerpiece of Mike’s technical approach is what he calls “orchestration” rather than coding. He’s using AI extensively to build the entire product, treating the AI more like a development team than a coding assistant.
Mike has developed a systematic methodology over months of iteration - one he considers core intellectual property and won’t share publicly. The approach involves extensive documentation, structured workflows, and quality controls that ensure AI-generated code meets production standards.
It took some effort to set up but the result of the effort? A substantial codebase built in four months, by one person who doesn’t write code manually. But Mike emphasized that the systems enabling this speed took significant investment to develop and refine.
Experience as Competitive Advantage
One of the evening’s most interesting tensions: Mike has never moved this fast technically, yet he’s more careful than ever about strategy. His previous startup failed because of “unforced errors” - things he now recognizes as avoidable with better judgment.
The coaching experience helps immensely. Having seen 1,000+ companies stumble on the same rocks means Mike and Jacqui (his co-founder, who has deep go-to-market expertise) share language and understanding. When Jacqui talks about what she needs for a marketing funnel, Mike knows exactly what to build and why. No translation layer. No miscommunication about priorities.
As Mike noted: “I think it’s atypical for your typical technical co-founder to understand go to market. And so because I actually did go to market at my last startup, because I’ve done a ton of coaching here on go to market and all those other pieces, Jacqui and I have a lot of shared language.”
This shared understanding enables speed: “There are days where I decided we’re going to use something and it’s going to cost us this much money to build, it’s like $19. And Jacqui’s like, okay. And it was done. That was the decision. We moved on.”
The Relentless Pace
But that speed comes with intensity. Mike and Jacqui work seven days a week. As Mike explained: “If we’re doing this, we are all in. We’re going hard.”
Why the urgency? Partly opportunity window. Partly momentum building momentum. And partly, as Mike admitted candidly: “I enjoy eating, and so I like to make money.”
The lesson from his previous startup - which ran out of money after a year - was clear: “I was frustrated and I only picked up the pace probably the last four months in that startup and started working seven day weeks, tried to save it, and at that point it was too late.”
This time: “I don’t want the fact that I didn’t work weekends for the first six months of the startup to be the reason that it wasn’t successful.”
The Hard Part: Customer Behavior Change
Despite all the technical sophistication and speed, Mike’s biggest worry isn’t competition or product execution - it’s finding customers and changing their behavior.
“Finding caregivers is really hard. They’re really busy. They don’t want to change their behavior.”
This is the classic challenge: building something people need versus building something people will actually use. The problem is real and painful, but getting people to adopt a new system when they’re already overwhelmed is the hard part.
Elderella is tackling the market directly to end users rather than through institutions - a harder path but one that keeps the focus on solving the actual user’s problem rather than the institution’s problem.
The bet: build something people will use in every context, not just one narrow use case. Make it indispensable by making it universal.
Mike’s current focus reflects this: “Jacqui and I are literally in our spare time right now reading psychology books” to understand how to drive that behavior change.
The Mistakes That Still Matter
For all the technical acceleration, the human challenges remain stubbornly difficult. Mike was blunt about the number one problem he sees in coaching: “Fighting with their co-founders, 100%.”
It’s so prevalent that during Mike’s time at Velocity, they hired two certified mediators specifically to help founders navigate co-founder conflicts.
His advice to founders facing relationship issues: “If you don’t fix this part of your relationship, if you don’t fix resentment, if you don’t do these things, this thing is going to get into a really nasty spot.”
Mike compared co-founder relationships to marriage, except worse: “You typically didn’t actually do a lot of dating before you got there. You are spending typically more time awake with this human than you do your actual partner. And at least if you get divorced, there are very clear rules. There are very unclear rules when startups break up.”
Building Systems, Not Just Product
One of the more striking aspects of Mike’s approach is how much infrastructure he’s built before having the core product fully validated. He’s not just building Elderella - he’s building the systems that make building Elderella sustainable.
Mike has created internal tools for documentation, testing, monitoring, and analysis - each designed to reduce friction and enable autonomous progress. The infrastructure prioritizes compliance and reliability from day one.
The lesson: invest in the systems that let you move fast sustainably, not just move fast once.
Core IP: The Skills Themselves
When asked if Mike shares his skills and agent setups publicly, his answer was firm: “No, and I never will.”
He and Jacqui consider this core intellectual property. “The amount of time and effort it took to make all of these things - eight or nine months.”
Mike draws the analogy to sales playbooks: “It’s just like a sales company. Here’s exactly what our script is for cold calling people, and here’s why it works, and here’s how we evolve that. If it’s working, then I’m not going to share it.”
His prediction: “I genuinely think that this is going to be the delta between 10x companies and 50x companies - the amount of effort and time that they’ve done to just completely hone” their AI development systems.
The Fundamentals Haven’t Changed
Despite the revolutionary speed of development, Mike kept returning to a theme: the fundamentals of building companies haven’t changed.
Go-to-market is still hard. Customer behavior change is still hard. Co-founder relationships are still the number one killer of startups. You still have to talk to users. You still have to understand the problem deeply. You still have to do things that don’t scale.
What’s changed is execution speed on the things that can be accelerated - primarily product development and internal tooling. But the human problems remain human problems.
When discussing his previous startup’s challenges, Mike listed them as “strategic errors” and “unforced errors” - not technical limitations. “It wasn’t the market, it wasn’t this. It was just not paying attention.”
The issues were common early-stage mistakes: team sizing, premature optimization of systems before product-market fit, and insufficient focus on validation. These mistakes are still possible - perhaps even easier - when you can build so quickly.
Key Insights for Founders
The evening crystallized several lessons:
Experience matters more when you can move faster - the ability to avoid known mistakes becomes the primary differentiator when technical execution is democratized.
Infrastructure investments compound - Mike’s internal tools, skills, and systems took months to develop but now provide sustained advantage.
The bottleneck has shifted from building to understanding - go-to-market, customer behavior change, and product-market fit are now the hardest problems, not technical feasibility.
Co-founder alignment remains critical - shared language and complementary expertise (technical + go-to-market) enable decision velocity.
AI development requires different discipline - context management, eval-driven development, and systematic knowledge capture replace traditional coding practices.
Relentlessness still matters - despite all the tooling, Mike and Jacqui are working seven days a week because momentum compounds and time matters.
As Mike put it when discussing his urgency: “The last startup that I did, we ran out of money after a year. We did not move fast enough. The lesson I took away from that is if it’s worth doing, I don’t want the fact that I didn’t work weekends for the first six months of the startup to be the reason that it wasn’t successful.”
The dinner wound down with founders pressing Mike for technical details, asking about prompt management, eval systems, and how to know if developers are using AI tools properly. Mike’s answer to the last question was characteristically direct: “Code review the PR. If the code review finds anything, they’re not using it properly.”
The standard, in Mike’s world, has shifted. Code reviews should find nothing because the AI should have already caught and fixed issues through multiple review cycles. Speed isn’t just about going fast - it’s about going fast and getting it right.
As builders finished their stew and conversations continued, the gap between theory and practice became clear. Many founders are experimenting with AI tools. Mike has systematized them into a competitive advantage.
The question hanging in the air: how many founders will invest the months required to build these systems when they could be shipping features? And how many will realize too late that the systems are the competitive advantage?
About this series
The Founder Dinner series is offered by Builders Club and Barn Ventures. The series features home cooked meals, founder discussion and connection, and is supported by a workshop series that has been developed with founders. This has evolved over nearly three years with the support of various people and organizations in the ecosystem. We have had a significant number of founders move on to raise capital from YC, Afore.vc, Garage VC, etc. Founders from companies like Datacurve.ai, Voltra, Automax.ai, and many others have attended.
Applications are open for the next series in the spring.