From Hackathon Winner to Y Combinator in 74 Days
On January 18, 2026, George Jefferson, Oliver Ulvebne, and Andrew Jefferson won the London Grokathon. Six weeks later, George, Oliver, and Art Freibrey took first place at the Mistral AI London Hackathon. By April 2, George and Art had been accepted into Y Combinator with Revnu.
The sequence unfolded in just 74 days. Yet the builders did more than collect prizes. Each hackathon gave them an opportunity to ship an ambitious AI product, demonstrate their technical ability in public, and strengthen the working relationships behind their next company.
Their projects were equally unconventional. Grok for Mayor searched more than 20 UK government APIs for waste and generated campaign videos. Mistralverse created a persistent AI civilization whose autonomous citizens worked, traded, governed, and elected their own mayor. Revnu later applied AI agents to a commercial problem: handling growth work such as advertising, SEO, and A/B testing.
Mistralverse was built at an Iterate hackathon and won first place in London. The team’s path shows how the right event can create more than a weekend prototype. A technical hackathon can become a proving ground for ideas, co-founders, and builders capable of launching a serious company.
From Two Hackathon Wins to Y Combinator
The path from the first prototype to Y Combinator took less than three months. Each milestone tested a different part of the team’s ability to build ambitious AI products under pressure.
| Date | Milestone | Project | Result |
|---|---|---|---|
| January 18, 2026 | London Grokathon by xAI | Grok for Mayor | 1st place |
| March 1, 2026 | Mistral AI London Hackathon | Mistralverse | 1st place |
| April 2, 2026 | Y Combinator acceptance | Revnu | George and Art accepted |
The projects changed, but the underlying pattern remained consistent: identify a difficult problem, build a functional AI system quickly, and explain the idea clearly enough to attract attention.
The two hackathons also created visible proof of execution. Grok for Mayor demonstrated the team’s ability to connect AI with fragmented public data. Mistralverse showed that its members could design a complex multi-agent environment around economic and political interactions.
Revnu then moved from experimental AI systems to a defined business problem. The company builds AI agents that perform growth work founders often postpone, including ads, SEO, and A/B testing.
This progression did not make hackathon success a guaranteed route into Y Combinator. It did, however, give the builders repeated opportunities to test their collaboration, produce tangible projects, and establish a public track record before launching their startup.
First Win: Grok for Mayor at the London Grokathon
On January 18, 2026, George Jefferson, Oliver Ulvebne, and Andrew Jefferson won first place at the London Grokathon organized around xAI technology.
Their project, Grok for Mayor, positioned an AI system as a DOGE-style watchdog for the UK. It searched more than 20 government APIs to identify potential public-sector waste, then turned its findings into campaign content for X.
The project combined several capabilities:
✅ Searching fragmented government data
✅ Identifying potentially wasteful expenditure
✅ Converting findings into accessible explanations
✅ Generating campaign videos
✅ Publishing the results through a recognizable AI persona
Grok for Mayor therefore went beyond placing a chatbot over public records. The team built a workflow that moved from data retrieval to analysis and public communication within one product.
The launch also attracted attention outside the event. George’s post presenting the project generated approximately 382,000 views on X, giving the team public evidence that an unusual AI concept could resonate beyond a judging panel.
Winning the Grokathon established the first proof point. The builders could form a team, execute quickly, and present a technically ambitious project with a clear narrative. Six weeks later, they repeated the result with a completely different system.
From Mistralverse to Y Combinator With Revnu
On April 2, 2026, just over one month after the Mistral AI hackathon, George Jefferson and Art Freibrey announced their acceptance into Y Combinator with Revnu.
Revnu addresses a familiar startup problem: founders enjoy building products, but growth work often remains fragmented, repetitive, or postponed. The company develops AI agents that execute tasks across areas such as:
👉 Paid advertising
👉 SEO
👉 A/B testing
👉 Growth experimentation
👉 Campaign execution
The startup was not a direct continuation of Mistralverse. The hackathon project simulated an autonomous AI society, while Revnu applies agents to practical growth operations. The connection lies in the builders’ ability to design autonomous systems, coordinate complex workflows, and turn an ambitious concept into a working product.
The timing is notable. George and Art moved from winning the Mistral hackathon to joining Y Combinator in approximately one month. Their two previous wins did not guarantee acceptance, but they provided concrete evidence of qualities relevant to an early-stage company: technical execution, rapid iteration, collaboration, and the ability to make an unfamiliar idea understandable.
Their progress also illustrates the value of building in public. Each project created a visible record of what the team could ship—not only a résumé claim or an untested startup concept. By the time Revnu entered Y Combinator, its founders had already demonstrated their ability to build and present original AI products under tight deadlines.
Why Hackathons Can Become a Founder Proving Ground?
A hackathon compresses several early-stage startup challenges into a short period. Builders must select a problem, limit the scope, divide responsibilities, ship a functional system, and persuade others that the result matters.
That environment reveals capabilities that a conventional application or technical interview may not capture.
1. Testing Co-Founder Compatibility
A promising idea cannot compensate for a team that struggles to make decisions or execute together. Hackathons allow potential co-founders to test how they communicate, divide work, manage pressure, and respond when the original plan fails.
George, Oliver, and Art did not only discuss building together. Their results provided evidence that they could produce a competitive AI system under a fixed deadline.
2. Turning Technical Skill Into a Working Product
Technical knowledge becomes more valuable when a builder can convert it into something usable. Both Grok for Mayor and Mistralverse required the teams to integrate multiple components around one coherent experience.
Judges could assess the output directly rather than relying on credentials or theoretical explanations.
3. Learning to Explain an Ambitious Idea
Hackathon teams have limited time to present the problem, demonstrate the product, and make the technical work understandable. This forces builders to develop a concise narrative around what they created and why it matters.
The same skill becomes essential when speaking with users, potential employees, investors, or accelerator partners.
4. Building a Public Track Record
Projects, rankings, demos, and social posts create visible evidence of execution. The attention generated by Grok for Mayor and the first-place result for Mistralverse gave the builders concrete work to point toward.
A hackathon win does not prove that a team can build a lasting company. It can, however, show that its members know how to move from an open-ended idea to a functional product quickly—one of the earliest tests every startup faces.
What Changed Across the Three Projects?
Grok for Mayor, Mistralverse, and Revnu serve different users and solve different problems. Together, however, they show a progression from time-boxed experiments to a company built around a repeatable commercial need.
| Project | Core concept | What it demonstrated |
|---|---|---|
| Grok for Mayor | AI system analyzing UK government data | Data integration, AI analysis, content generation, and public storytelling |
| Mistralverse | Persistent society of autonomous Mistral agents | Multi-agent coordination, shared state, economic behavior, and governance |
| Revnu | AI agents executing startup growth work | Application of agentic systems to recurring business operations |
The first two projects were designed for hackathon environments. They needed to capture attention quickly, work within a limited development window, and demonstrate a technically original idea.
Revnu required a different shift. Instead of optimizing for one memorable demo, George and Art selected a recurring pain point: founders need growth, but often lack the time or desire to manage every campaign and experiment manually.
The technical ambition remained, but the product logic became more commercial. The focus moved from showing what autonomous agents could do to using them for work customers may pay to automate.
This is an important distinction. Hackathon projects do not need to become startups to create long-term value. Their methods, team dynamics, technical lessons, or individual builders can carry into a completely different company. In this case, the winning projects became evidence of execution before Revnu emerged as the business.
Build and Prove AI Ideas With Iterate!
Iterate runs technical hackathons where ambitious builders turn difficult AI problems into working products. Each event gives participants a defined environment to test ideas, form teams, and demonstrate what they can build under real constraints.
Mistralverse emerged from an Iterate hackathon and won first place in London. The project gave George, Oliver, and Art a public, verifiable example of their ability to design and ship a complex multi-agent system.
Iterate manages the complete event:
- Challenge definition
- AI builder sourcing
- Application review and team formation
- Technical mentorship
- Event operations
- Project evaluation
- Final demos and result documentation
- Identification of high-performing builders
For AI companies, the format produces working prototypes and direct access to technical talent. For builders, it creates an opportunity to test potential teammates, develop original systems, and build a public record of execution.
Run an AI hackathon with Iterate to transform a technical challenge into working projects built by proven AI talent! ⚡
Conclusion
Two hackathon wins did not automatically lead to Y Combinator, but they gave George, Oliver, and Art repeated opportunities to build together and prove their execution skills. Their journey shows how technical hackathons can turn ambitious ideas into working products, public results, and potentially lasting founder relationships.
Frequently Asked Questions
Can winning a hackathon help founders enter Y Combinator?
A hackathon win does not guarantee acceptance into Y Combinator. However, it can provide evidence that a team builds quickly, collaborates under pressure, and turns ambitious ideas into functional products.
Did Mistralverse become Revnu?
No. Mistralverse is a persistent civilization of autonomous AI agents, while Revnu develops AI agents for growth activities such as advertising, SEO, and A/B testing. George and Art built both projects, but they serve different purposes.
What can founders gain from an AI hackathon?
Founders can test potential teammates, validate technical ideas, receive feedback, create a working prototype, and establish a public track record. Companies can also use hackathons to identify skilled AI builders through real project work.
How does Iterate organize AI hackathons?
Iterate handles challenge definition, builder sourcing, applications, team formation, event operations, technical evaluation, and final demos. Each event connects difficult AI problems with builders capable of developing and presenting working solutions.