LLMs Ate the Search Bar, Now What?

Max Sinclair blew our minds last week ( Hendrik Laubscher ) and I. We welcomed him onto the Ostrich Report to have what we thought would be a normal enough discussion about AI and its productivity gains for markketplace sellers. Wrong. Azoma occupy a unique space and have a long history here too.

2 co founders with skills and experience, one from the Death Star (Amazon) who combined to build products for customers such as Mars - their customer list almost seems secondary to them - the product is critical. Isn’t everyones? No. Nor is their product, for evereyone I mean. His words, not mine - refreshingly honest and clear on their purpose. We even talked ethics and controls needed to run in parallel with AI growth. Max argues for user-controlled “reset my algorithm” and a hard line against government data centralisation.

Azoma have 2 patents - 1 granted and 1 pending. This in itself, in this field is incredible. But it tells a tale of masters of their craft.

I first met Max where we shared a gin in Belfast overlooking Harland and Wolf, that infamous dockyard.Max was confident and quietly assessing the room. With hindsight I can see he had already outgrown most of us.

We talk: answer engines, agentic browsers, and soon physical AI will reshape discovery, ads, and the P&L politics inside brands. .

Highlights

From keywords to conversations: LLMs are already baked into Amazon, Walmart, and others — whether you see the chat UI or not. “Search” is becoming ask → answer → act.

Azoma’s patents underpin a system that simulates how people talk to AI engines at scale, tracks citations/crawler patterns, and models brand share of voice in AI.

Who’s buying this stuff: Pilots start with central/search CoEs, but brand teams fund the roll-outs. Which is a budgetary shift.

Ads vs answers: In an answer-first world, users won’t tolerate ad clutter. Expect new monetisation (affiliate/referral rails, Stripe-like takes) and hyper-personalised, generated promotions, not today’s slotting.

What’s next (near-term): Agentic browsers, then true multi-step agents, then physical AI (smart fridges, mirrors, in-home devices) and lightweight AR moments (hello, Ray-Ban Meta).

Ethics with teeth: The dangerous AI wasn’t GenAI; it was the deterministic engagement algorithms we couldn’t reset. Max argues for user-controlled “reset my algorithm” and a hard line against government data centralisation.

5 Big Takeaways for Operators
AEO > SEO: Start treating AI engines as distribution. Track your share of voice in AI, your citations, and how prompts/personas surface (or bury) your brand.
Move the budget.

Design for questions, not keywords: Your product data, FAQs, UGC, and how-to context must answer situations (“desk has a drawer; need clamp”), that’s what LLMs reward. More in the pod.

This was our best yet.

Someone should sponsor this. Are you listening Rithum

Episode here: https://lnkd.in/eexBn3KM