The GEO category was built on a simple insight: AI is changing how people find things. Companies like Profound raised $96M to prove that insight. Scrunch raised $15M. Others followed.

Johannes Erett watched all of this and made a different bet.

LunaLift entered later, with less capital, and stayed invite-only. When the South Summit jury picked them as the Digital & Tech winner, over companies with a significant head start and far more funding, Johannes had a clear read on why.

“The signal I trust more than the award itself is that we’re already winning customers away from some of the largest players in the category. Particularly sophisticated companies and agencies that have reached the limits of first-generation tools and need something more powerful.”

The first wave proved there was demand. The next wave will be defined by who can deliver the best outcomes.

Most GEO platforms today are visibility trackers. They tell you where your brand appears in ChatGPT or Gemini. That was useful in 2024. Buyers are asking different questions now.

“Not just where do I show up in ChatGPT, but how do I improve it, how do I influence it, and ultimately, how do I generate revenue from it?”

That’s where Johannes thinks the market is starting to separate.

“Enterprise buyers don’t ultimately pay for monitoring. They pay for outcomes.”

LunaLift built attribution early, before most competitors thought it mattered. The question they built for was not only whether a brand appears in AI search, but what happens after it does. Indirect AI influence. Customer journeys across channels. The difference between a mention and a conversion.

“Companies don’t ultimately care about rankings or visibility scores. They care about customers, conversions, revenue and growth.”

Why lean was a deliberate choice

There is a common assumption in startups: distribution beats product quality, especially when AI makes it easier to build and copy software.

Johannes disagrees. With a specific caveat.

“That’s only true in markets where the product itself doesn’t matter very much.”

He uses an analogy that lands differently than the usual startup metaphors.

“Walk onto a professional construction site, a manufacturing facility, a Formula One garage, or an operating room and you rarely find people relying on the cheapest or weakest tools available. The cost of poor performance is simply too high.”

His read on software is the same. Early in a category, distribution often wins because buyers are still learning what matters. As the market matures, they stop buying the story and start evaluating outcomes.

The data he cites is specific: nearly 70% of users stop using a product within three months of signing up. Acquisition creates categories. Retention determines winners.

“AI is making average easier. It is not making excellence easier.”

The invite-only access fits the same logic. LunaLift is deliberately working with a selected group of partners, some of the best marketers in the world, rather than opening the doors to everyone at once. Johannes is direct about why: close collaboration with a small, exceptional group lets the team move faster and build deeper than a broad rollout would, where focus gets split between scaling and refining the product.

“We can go much further, much faster, by building alongside a few of the best than by spreading ourselves across everyone at once. The waitlist isn’t scarcity for its own sake. It’s how we keep that focus.”

What the scientific background actually changes

Johannes holds 7 US ML patents. He scaled Cognism’s ML team pre-IPO. His background spans nanoscale engineering, robotics, lab automation.

Ask him where that shows up in LunaLift and the answer is not what you’d expect.

“Less in the specific technologies and more in how we think.”

In robotics and lab automation, reality exposes weak assumptions very fast. Either the system works under changing conditions or it doesn’t. That creates, in his words, a mindset of precision, feedback loops and first-principles thinking.

“You learn to care about what is actually happening rather than what sounds convincing.”

He applies that to the category broadly. Many GEO companies start from marketing assumptions and work downward. LunaLift starts from how these systems actually function and builds upward.

On the ground truth that keeps shifting

The obvious question for any GEO company is stability. LLMs update constantly. What’s trusted today can disappear in the next training run. How do you build something durable on that?

Johannes pushes back on the premise.

“The biggest misconception in this space is that everything changes every few months.”

There are layers to these systems. SEO is not dead. Google AI still relies heavily on Google Search. ChatGPT still relies heavily on Bing. Backlinks, authority and reputation still matter because AI systems need reliable sources to ground their answers.

At the same time, there are genuinely new dynamics. Models breaking prompts into multiple searches, evaluating sources, calling tools, synthesising answers, and increasingly acting on behalf of the user in ways traditional search never did. He cites a specific number: ChatGPT citations overlap only around 12% with Google’s results. That gap is where the opportunity lives.

“We don’t see model updates as a risk but as the new normal. Our product is trained to adapt to them automatically.”

What he’s actually building

Where most of the category stops at tracking, LunaLift runs the full cycle: strategy, content sophisticated enough to carry a brand’s voice, the company’s own experts and AI agents working side by side, and attribution that ties it back to real conversions rather than a visibility score.

On who it’s for, Johannes pushes back on the idea that being selective means staying small.

“There are millions of companies creating genuine value, solving real problems and earning trust through the quality of what they do. Those are the companies we want. That’s not a niche.”

The selectivity comes from how the product works. LunaLift provides the services AI systems rely on to verify that content is authentic and where it came from, called many times a second as AI search runs. Putting its name behind that verification means its incentives are tied to keeping AI search trustworthy.

“So yes, we’d turn customers away. If a company wants to manufacture trust rather than earn it, we’re not interested. The good news is that there’s no shortage of companies worth working with.”

Johannes believes the businesses that win won’t be gaming individual models. They’ll be building genuine expertise in their fields and using systems that translate that expertise into trust, visibility and influence as AI search evolves.

“In the long run, I believe the best products usually win because excellence is difficult to copy, while hype eventually runs out of road.”

Johannes Erett is CEO and Chief Scientist at LunaLift. We spoke with him at South Summit Madrid 2026, where LunaLift won the Digital & Tech Solutions category. The Conversion Media publishes interviews with B2B founders and revenue leaders. If you’d like to participate, get in touch.