For two decades, the playbook for brand discovery was stable. Buyers searched on Google. Brands optimized to rank. Pages competed against pages, and the winners earned the click.

That model is unwinding faster than most marketing teams have internalized. A growing share of buyer questions never reach a search engine at all. They get answered by an AI assistant — directly, in conversation, with one or two cited sources rather than ten blue links.

This isn’t a new channel to add. It’s a structural shift in the substrate of discovery.

What’s actually changing

The tactical version of this story — “you need to optimize for ChatGPT” — is too small. The category most agencies are currently selling under labels like AEO and GEO is real, but it’s only the first layer of a much longer transformation.

We see four layers stacking up over the next decade.

The citation layer is the one most teams are starting to think about. It’s where AI assistants surface specific brands when answering buyer questions, and where the work today centers on becoming a source those systems trust.

The agent layer is next. AI agents will increasingly take action on behalf of buyers — not just answering questions but completing tasks, comparing options, making recommendations, and eventually making purchases. Brands that aren’t legible to those agents won’t get considered.

The infrastructure layer comes after that. Brands themselves will need to exist as structured entities that AI systems can represent — knowledge graphs, capability descriptions, real-time data feeds, agent-readable APIs. The brand becomes a piece of data infrastructure, not just a story.

The market layer is the longest horizon. Agent-to-agent commerce, AI-mediated marketplaces, dynamic pricing and matching that happens between machines representing buyers and sellers. The brand exists primarily as a counterparty to other AI systems.

Why most of what’s being sold today is too tactical

Most agencies operating in this space are selling pieces of layer one. Schema markup. FAQ optimization. Reddit comment seeding. AI Overview citation guarantees.

Some of that work is real and useful. A lot of it is SEO vocabulary translated into LLM-shaped marketing copy. And almost none of it prepares a brand for layers two through four.

The harder, more durable work is building the kind of brand authority that compounds across all four layers — credibility signals AI systems are trained to recognize and trust at every level of mediation. Founder-led video that produces high-authority transcripts AI can parse. Genuine community presence in the surfaces AI cites most. Editorial coverage on domains AI treats as primary sources. Entity infrastructure that survives every model update.

That’s a much bigger project than ranking in ChatGPT this quarter. It’s also the only project that’ll still matter in five years.

What this means for brands now

The window for being early is open. Most brands haven’t started. The ones that begin now — building real authority across the surfaces AI retrieves from, with founder voice as the anchor — will compound for years before the category catches up.

The brands that wait will spend the next decade trying to buy back authority they could have built for a fraction of the cost.

We started Craneweave to do this work for ambitious brands willing to think a layer deeper than tactics. More to come.