I spent the last week in Malaysia participating in an AI workshop and hackathon with one of my clients. A few things stood out, the biggest had nothing to do with AI search, and everything to do with context and a knowledge base.

I opened with a talk on how I actually use AI day to day as a solo founder (below). Connecting MCPs, running tasks to help power my marketing, cutting the noise so I only act on what matters, building Obsero.

But the real lesson came from the work itself.

Generating content with AI is the easy bit. Everyone can do it now. A prompt and 10 seconds gets you a post, an article, a campaign.

The hard bit, and the bit with actual value, is the intelligence that sits in front of the generation. Knowing what's worth making at all.

That means feeding the model the things only your business knows. What's moving in your market right now. What competitors are doing. What's already been covered. What fits your brand and what doesn't. Then letting it tell you what to make, and why, before it makes a single thing.

If you're building AI into your business, that's where I'd put the effort. Not a better prompt for writing posts. A layer that decides which post is worth writing.

Same goes for AI search. Showing up in ChatGPT or Google's AI answers isn't a content volume problem. It's a "do you actually know what to create, and where the gaps are" problem. Generation without that intelligence just makes more noise and generic outputs, and the models are already drowning in it.

The teams that win with AI won't be the ones generating the most. They'll be the ones who built the judgement layer.

Feel free to download an extract of my slides below.

AI has changed what visibility means. It is no longer about where you rank in ten blue links. It is about whether AI associates your brand with the subjects that matter in your market, and how strongly. That association is what decides whether you get named in an answer. The shift is that this can now be measured directly, for the first time.

For many years, the industry focused on keywords. Does the page use the right terms, in the right places, enough times. That made sense when the job was ranking a page in a list of links. It does not work for AI, because AI does not look at ‘keywords’. It looks for meaning. A page can use every right word and still be only weakly associated with the subject, because association is built on meaning, not vocabulary.

Keywords tell you whether the words are present. Meaning tells you whether the page is genuinely about the subject. Intent tells you whether it answers what the person actually wanted. Old audits stop at the first. AI judges you on the last.

To show what this looks like in practice, I ran it on a real brand. It is a name you would know, so it stays anonymous here, but the category does the explaining: this is a market where people ask AI things like which heels suit a wedding, what loafers feel timeless, which sandals work for a holiday. The charts below are that brand's actual results. Every point is one of those subjects, and where it sits tells you how the brand is doing on it.

This is where vector embeddings come in. An embedding is a way of turning a piece of text, a page or a question, into a string of numbers that captures its meaning.

Put your content and the subjects your market cares about into the same space and you can measure how close they sit.

Close means your content is strongly associated with that subject. Far means it is not, whatever the page says on the surface.

Do that across every subject in your category and you get an objective map of your content strength. Where you are strongly associated and likely to be surfaced. Where you are thin. Where you are absent entirely.

Association strength tells you what your own content earns you. It does not tell you the whole story, because AI does not build its answer from your site alone. It assembles each answer from across the web: your pages, editorial, retailers, forums and more. The citation data lets you see that makeup.

This is the part that surprises people. A brand can be strongly associated with a subject in AI answers while almost none of that association comes from its own content. One brand we looked at was named in roughly a quarter of AI answers in its category. Its own pages were the source just 1.5% of the time.

So the citation layer answers a different and equally strategic question. Not how strong is my content, but how much of my position do I actually own, and where is the rest coming from.

Obsero lets you pull the two together for a complete read of where you stand in AI.

The first tells you whether your content is strong enough to be associated with the subjects that matter.

The second tells you how much of that standing you own, versus how much is assembled from everyone else. One is the strength of your content. The other is the source of your visibility. You need both to know where to invest.

This is the new content audit, and the data is sitting in your Obsero toolkit. But it should not be a one-off snapshot. Once you have made the changes, measure again and check your association with the entity is actually strengthening, not just assume it is.

Want to see it in action? Reply to this email or drop a comment below, and I'll do a video walking through it in Obsero

That is all for this week. Until next time.

Please do let me know what you thought of this edition - I promise to respond.

Andy