Lumian Gen AI Newsletter Issue #62

Welcome to the 62nd edition of the Lumian Gen AI Newsletter!

For two decades, “search” meant Google. You typed a few words, got a ranked list, and someone in Mountain View quietly decided your fate. Search Engine Optimization (SEO) became a cottage industry: backlinks, metadata, keyword stuffing, rinse, repeat. Then came ChatGPT, Gemini, Claude and suddenly, users stopped clicking links and started talking to essays. 

Now there’s a new acronym: GEO (Generative Engine Optimization). Or LLMO. Or AIO. We haven’t agreed on a name yet, but the point is clear: the answer box is eating the search results.

In the old world, visibility meant showing up first. Now, it means being the answer. Models don’t show links; they summarize, synthesize, and hallucinate with confidence. GEO is the idea that you can somehow influence what these models say and optimize not for clicks, but citations. Except we don’t really know how. Unlike Google, LLMs don’t publish rules. What shows up today might disappear tomorrow.

Naturally, a cottage industry of tools has emerged. Some track mentions in model outputs. Others claim to optimize content for “model memory.” But most rely on APIs, not the actual UI users see. That’s a problem. Users engage with interfaces, not raw models. So GEO dashboards may be measuring the wrong thing entirely.

This is the difference between observing and influencing and right now, we’re mostly observing. Still, if GEO becomes real and if models start reliably quoting brands, the upside is massive. The $80B SEO industry is definitely evolving and brands will want to know: are we remembered? Are we referenced? Are we trusted? And maybe more importantly: can we buy our way in? 

Unlikely. Unlike Google, OpenAI and Anthropic don’t need third-party content to make money. They sell subscriptions, not traffic. Which means you can’t bid on keywords and you have to earn relevance.

That’s a branding problem, not a performance one.

GEO’s promise is huge: influence what the model says, track the results, close the loop. A new channel, driven by prompts instead of pixels. But it’s early. The rules shift constantly. Interfaces change. What works today breaks tomorrow. There’s no stable surface to optimize against.

Still, people built massive companies on SEO without knowing Google’s algorithm. Maybe the same will happen with GEO. And that’s the core question. 

GEO is a mindset. If language models are the new front door to discovery, you’re not trying to rank, you’re trying to stick in the model’s head. If it forgets you, you disappear.

So yes, GEO might matter. But only if you matter first.

Happy reading! 📚🤖🎵

In this week’s issue:

• News Flash: Google I/O, Anthropic’s Claude 4, OpenAI’s io

• AI in Practice: AI Tools I’m Tinkering With

• Fundraising: The biggest deals in AI

• Nerd Out: Technical and Business Content for Everyone

🚀 AI in Practice

AI Tools I’m Tinkering With

• Experience by Odyssey - AI video you can watch and interact with

• Snapdeck - AI slide generation in Figma