For this edition of the Studio Roadshow, we sat down with Mstudio, a venture studio operating across West Africa (Brussels-based, HQ in Côte d'Ivoire) that builds companies for emerging markets with replicable business models across Africa. They've funded 68% of all startups that raised money in Côte d'Ivoire over the last three years. That's a real signal about what market depth looks like when you build with local knowledge.

The topic was supposed to be simple

How do you use AI to support the studio process and the startup itself?

What it turned into was a more uncomfortable conversation: what actually makes a venture studio defensible when every founder on Earth now has access to the same AI tools you do?

Mstudio has built 100+ AI agents across their entire playbook: from ideation through exit. Builders has been building a unified data and memory layer that replaces stage-gate models altogether. Different geographies, different approaches, same underlying conclusion.

Where it got interesting

Mstudio: document everything, automate everything

Mstudio took their entire venture-building playbook: 100+ multidisciplinary tasks built over three years, and turned it into a system of AI agents. The logic is clean: if you can document it, you can automate it. They cover everything from ideation and customer discovery to legal review, grant scanning, VC preparation, and M&A acquirer mapping.

A few things stood out.

Their EIR programme is fully sequenced into numbered agents that map 1:1 to weekly playbook tasks. An entrepreneur in residence doesn't just get a mentor, they get a structured, AI-augmented process where every task has a corresponding agent. One key learning from Mstudio: keep scope narrow. Asking an agent to do "mom and dad at the same time" doesn't work. Split the task. Let humans approve between steps.

They've also built agents we hadn't seen before. A system that scrapes an investor's public statements, podcasts, and investment history, then briefs the founder before a pitch call on what that specific VC cares about. A comparable-company watchdog that sends regular reports on what competitor startups in other emerging markets are doing. A grant scanner that monitors funding sources and automatically matches them to portfolio companies. An M&A agent that maps potential acquirers years before a startup is ready to exit: not to sell, but to build the relationship early.

The output is consistent. Same format, same structure, every time. That consistency is the point. When Mstudio's co-founder tested a new agentic AI assistant, his complaint wasn't capability, it was that asking for the same output three times gave three different results. Their agents don't do that. The skill is in the structure.

Builders: replace the playbook with data

We went a different direction. Instead of automating a playbook, we replaced it.

Every conversation a founder has: discovery calls, investor meetings, sales conversations gets recorded, transcribed, and processed through our system. We extract signals: pain points, workflows, competitors, objections, buying signals. Each signal gets an evidence level from E1 to E5 (quantified, story-driven, ROI-backed). And signals decay: a pain point discussed six months ago that never resurfaces carries less weight than one that comes up across multiple conversations.

This changes the fundamental studio interaction. We don't ask founders "how's it going?" anymore. The system already knows how far along the sales pipeline is, how confident we should be, and where things are stuck. What we showed Mstudio live: auto-generated problem one-pagers, stakeholder maps, validation gaps, competitor maps, pivot histories, all from 955+ signals extracted across a company's lifespan. No founder had to write a single document.

The hardest part wasn't building this. It was resisting the urge to add more. Our engineers built a chat interface on top out of excitement: we removed it. What founders need isn't an open-ended AI assistant. It's structured, high-quality data connected to the decisions that actually matter.

Where we clashed, and where we agreed

Agents aren't the moat. Data is.

This was the real turning point. In a world where you can build 100 agents in six months, where there are agents that build other agents: What's actually defensible?

Mstudio's advantage isn't that they have 100 agents. It's that they have three years of documented processes, deep localised knowledge about Francophone African markets, and a playbook tested across 17 ventures. The agents are the delivery mechanism. The data and context behind them is the moat.

The same logic applies on our side, differently. We care about temporal context, not just what happened, but when, who decided it, what changed as a result, and whether it matters more or less today than last month. Every signal in our system has a timestamp, an author, and a decay function.

The studio with the best data wins. The studio with the fanciest agent loses.

More tools, more noise

Here's something the AI conversation tends to skip over: connecting more data sources often makes things worse, not better. Founders' Notion pages, meeting notes, AI-generated documents, all of it creates bloat. Ten versions of a value proposition doc, no clear source of truth, and AI makes it faster to generate more of it.

Mstudio faces a related but different challenge. Their structured agent outputs solve the format problem, but they still deal with founders not producing enough data: particularly in West Africa, where a significant portion of conversations happen offline and go unrecorded.

Both studios agree: the answer isn't more data. It's better data with clearer structure.

On founders who don't engage

Not all founders adopt AI tools equally. Mstudio has portfolio companies that simply aren't AI-minded enough to use the agents effectively. If a founder in 2026 won't engage with AI-augmented processes, that's a signal about the founder, not the tool. The bar is moving, and studios need to decide what side of it they're on.

What both studios are actually saying

Strip away the tools, the agents, the dashboards. Both Mstudio and Builders are making the same bet: that venture studios survive by being honest faster than everyone else.

Mstudio does this through rigorous documentation and automation: Every task structured, every output consistent, every comparable tracked. Builders does this by making the data tell the truth whether the founder wants to hear it or not, surfacing where someone pivoted away from a hard customer question, and whether that pattern is showing up across hundreds of conversations.

The approach is less important than the principle: systems should show you the truth faster, not hide it behind pretty interfaces.

What we're taking forward

For Mstudio, the opportunity is feedback loops. Their agents produce excellent structured output, but the connection between what agents recommend and what actually works in the field, particularly in markets where most interactions happen offline, is the next frontier.

For Builders, the validation is clear. Our system captures the kind of signals Mstudio is automating, but without requiring the founder to do anything. The combination, structured process and passive data capture, is where both studios see the future.

And for anyone building a studio in 2026: if your data is messy and your process is undocumented, no amount of AI will save you. But if you've done the hard work, and both studios in this room have, AI doesn't just help. It forces you to confront what's actually true about your portfolio, your founders, and yourself.

That's either terrifying or exciting. We think it's both.

If your studio is working through similar questions about automation, data, and founder transparency, we'd like to compare notes.

This is the 4th edition of the Studio Roadshow. Catch up on previous editions with High Alpha, Polymath Ventures, and OSS Ventures.