How a 100km run inspired a mission-led tech startup: Ben Burch of Allegr

Ben Burch set out to run 100 kilometres and expected friends to join him for five or ten each. Instead, people who turned up for five ran fifteen, most of them had never met, and something happened over the course of that day that he has spent the years since trying to rebuild deliberately. The result is Allegr, a community platform built on a deliberately unfashionable idea: that the point of community software is to get people off their phones and into the same room. In this episode Ben talks to Rob Woodhead about the personal reason he started it, why social media is structurally incapable of building community, how Allegr found its first real users among military veterans, and what two decades in technology have taught him about projects that quietly consume weeks. He is also candid about where generative AI helps and where it simply relocates the problem.

Key takeaways

• Social platforms optimise for market building rather than community building, so their incentives run directly against people actually meeting up.
• Communities form around three things: a shared identity, a common purpose, and a transition that destabilises both.
• Veteran communities are underserved by technology because the landscape is regimental and closed rather than commercially open.
• Depth of impact on a small number of people is a more honest measure of success than shallow engagement across many.
• Generative AI delivers roughly the first 80% of a build very quickly, then the final 20% consumes most of the time it appeared to save.
• Developers who understand why they are building something make fewer damaging assumptions than developers handed a ticket.
• Framework and tooling churn is fast enough that every architecture decision should be treated as temporary.

Episode chapters

• 00:26 From Oxford rowing to building a business
• 03:49 The personal story behind Allegr
• 06:11 What a 100km charity run revealed about community
• 09:35 Why technology had to be the catalyst
• 10:44 Why social media is not social
• 13:03 Finding an underserved community
• 17:04 How shared identity creates instant trust
• 21:51 Lessons from years of technical consulting
• 26:24 The mistakes that cost weeks
• 30:30 Where generative AI still falls short
• 32:13 Where Allegr is now
• 39:55 Expanding into sports and alumni communities
• 43:44 What it is like working with an external development team
• 50:31 Why process is not the point

Why is social media bad at building real communities?

Because its goals are not aligned with yours. Ben is blunt about it: social media is an advertising platform, and every metric, action and piece of engagement inside it is designed for market building rather than community building.

That distinction matters more than it first sounds. A platform funded by attention has no commercial reason to help you leave it. If two people meet up in person, that interaction happens somewhere the platform cannot measure or monetise, which makes it worth less to the business than the same two people arguing in a comment thread.

Allegr had to be built differently from the ground up rather than as a friendlier version of the same thing. Ben is clear that a surface-level difference would not have survived contact with commercial pressure. Without strong positions on data, monetisation and purpose baked into the product early, the pull towards the standard model would eventually win.

He also points at a contradiction that a lot of organisations have not noticed. Plenty of regimental associations and alumni groups run their communities through Facebook while simultaneously campaigning on wellbeing. If the tool you use to support people is one you accept is not a net positive for them, the support and the mechanism are working against each other.

What actually holds a community together?

Three ingredients, according to Ben, and none of them are features.

The first is a shared identity. The second is a common purpose. The third is the one most people miss: a transition. Communities matter most to people at the moment their identity is shifting. Soldier to veteran. Athlete to former athlete. Adult to parent. Those transition points are genuinely destabilising, and they are exactly when belonging stops being a nice-to-have.

Ben found this out by accident. He organised a 100km charity run and expected friends to join him for five or ten kilometres each. What happened instead was that people who came to run five ran fifteen, people who came for ten ran thirty, and most of them did not know each other. There was a reason to be there, a shared experience, and a common purpose, and that turned out to be enough. The run itself was not the thing worth repeating. The conditions that made it work were.

When the team later mapped these traits against different kinds of group, the pattern held in both directions. Sports clubs, schools, regiments, universities, all of them have identity, purpose and transition in some proportion.

Why are veteran communities hard to serve with technology?

Partly because the landscape is closed. The UK veteran world is organised regimentally rather than commercially, which means it is not a market you can simply advertise into. Allegr got in because Robin, who joined the founding team, had served and could open the door to his old regiment.

Partly because of the demographic. Ben is realistic that many veterans are older and that persuading them to use technology to do things offline is not low-hanging fruit.

And partly because the outcome is hard to measure. Allegr is explicit that it is not a crisis service and not a mental health platform. The bet is upstream: if people keep talking to each other and keep turning up to things, fewer of them reach crisis in the first place. Ben would rather make a deep difference to five people than change the morning routine of fifty, which is a much harder number to put in a deck.

What makes it work at all is instant trust. If someone served in the same regiment, you may never have met them, but there is enough common ground to hang a conversation on. That small hook is the difference between an invitation being ignored and someone showing up.

It is the same pattern we see in the platforms we build for tightly defined user groups. You can see how we have approached projects like this across fintech, healthcare and high-scale events.

Why does generative AI struggle with the last 20% of a build?

Because the hard part was never the first 80%.

Rob puts the observation to Ben directly: AI tools get you a long way very quickly, and then debugging what the machine wrote turns out to be genuinely difficult. All the time you saved goes into making the thing actually work. Ben agrees, and he does not think it is a new problem. It is the same shape as the oldest problem in software delivery, where you are perpetually ten minutes from finishing and those ten minutes take ten hours.

His warning is about time drain rather than about AI specifically. Technology consumes enormous amounts of time precisely because it feels close to done. He tells a story against himself: years ago he spent nights and a serious amount of money building clusters of Exchange servers for clients, proud of the engineering, right up until Microsoft launched Office 365 and the whole thing became worthless overnight. The engineering was fine. The looking-up was not.

He also thinks the industry made this worse for itself, stacking framework on framework and abstraction on abstraction until the mess became structural.

This is the gap our Vibe Code Audit exists to close. If an AI-assisted codebase is 80% of the way there, the useful question is what the remaining 20% will actually cost you in security, scalability and maintenance. You can also read more about our AI software development work.

What is it like working with an external development team?

Ben went into it with his eyes open, having worked with internal teams, external teams, local teams and remote teams before. Old St Labs designed and built the Allegr platform, with the delivery team based in the Philippines and the client relationship held in London.

Two things made it work for him. The first was continuity: the people who started the project are the people still on it. The second was the local point of contact, which gave him somewhere to put trust while the build was happening elsewhere.

He is even-handed about the friction. There are differences and challenges, and he thinks both sides own them. But the one he singles out is not about distance at all. Developers, in his experience, will hear that something is broken, dive straight in, fix it, and report back only once it is done. That can feel like silence. It is not a remote-team problem or an agency problem. It is the temperament of people who like solving problems, which is exactly who you want building your product.

You can meet our engineering team in the Philippines and read more about how Old St Labs works.

Should developers talk to clients directly?

Ben thinks yes, and he has a specific reason for it.

If a developer does not know why they are writing something, they will make assumptions. Small ones, mostly, but assumptions accumulate and the product that arrives at the other end is quietly worse for reasons nobody can point at. Context is not a nice extra for the person writing the code. It is an input to the code.

He has acted on this before. Running his own teams, he removed QA from the process deliberately, so that developers had to feel the impact of a bug rather than watching it become somebody else's ticket. On the Allegr project, what he valued was developers being on the calls and hearing requirements firsthand rather than through a chain of translation.

Rob agrees, with the caveat that it is a permanent work in progress. His own role is often to sit in the middle and ask the developer to explain the problem in plainer language, or ask the client to explain the requirement in more concrete terms.

Ben's wider point is that the technology industry is too attached to process for its own sake. Give him a few developers who genuinely understand what they are building and he thinks they can build almost anything, because people who understand the goal ask better questions than people following a workflow.

About the guest

Ben Burch is co-founder of Allegr, a community platform used by UK military regiments, veteran organisations and sports clubs, and a Technical Consultant at Old St Labs. A former Oxford University Boat Club president and national squad rower, he has spent two decades working across e-commerce, financial technology and technical consulting before building his own business.

About the host

Rob Woodhead is a co-founder of Old.St Labs, a London-based AI software agency building web and mobile products for startups and scale-ups. He hosts The One About AI alongside Rob Lynch.

‍

‍

‍