There are few people better placed to help us understand the AI revolution than Dr. Daniel Hulme, who is now the Chief AI Officer at WPP. Daniel is one of the few people actually qualified to separate the AI signal from the noise. Here are a few takeaways from today's podcast interview.

There’s a lot of chatter that AI will make Software as a Service obsolete. Daniel’s take? "This is completely wrong". We aren't seeing the death of SaaS; we’re seeing a total architectural rewrite.

Traditionally, SaaS has been a three-layer cake: a database, some business logic, and a UI. AI is absolutely demolishing the UI layer. Whether it’s voice, AR glasses, or LLM-driven interfaces, the way we talk to our data is changing forever.

But here’s the kicker for founders: don’t start with the AI. Start with the problem. Sometimes a simple macro or a traditional optimization algorithm is better (and cheaper) than a licensed LLM.

We are currently in a cycle of massive overestimation. Daniel describes today’s AI as an “intoxicated graduate"—it’s smart, it’s fast, but it lacks reasoning and often hallucinates its own reality.

To win, you have to stop chasing low-hanging fruit. If a third-party vendor can solve it for a few bucks, it’s not a moat. The real value lies in the gnarly supply chain problems. And stop waiting for your data to be ready—it never will be. Start with the problem and force the data to catch up.

Forget SEO. The new frontier is GSO.

Historically, brands used ads to build associations in the human brain—like Volvo and "safety". In the AI age, you have to build those same associations in the LLM’s "brain". GSO is about ensuring that when a user asks an AI for a recommendation, your product is the one the model thinks of first.

We are also seeing the rise of AI agents with purchasing power. We’re moving away from static customer segments toward dynamic segment identification. Your next big customer might not even be a human.

Daniel isn't worried about AI destroying jobs in the next five years—he expects an explosion of innovation. But in the long term, we are heading toward an "Economic Singularity".

By removing the friction of human labor, we could reach a state of abundance where the cost of food, healthcare, and education drops to near zero. This isn't just about efficiency; it’s about freeing people to live their true humanity.

Walmart (February 19, Pre-market): Investors eye e-commerce growth and retail dominance.

Be Careful What You Prompt For

The "Summarize with AI" button is the new Trojan horse, and frankly, we should have seen it coming. For decades, the SEO industry has been a game of cat-and-mouse—gaming algorithms to climb the search rankings. But as we pivot from search engines to AI assistants, the battlefield has shifted from the "Page 1" results to the "Memory" of your digital co-pilot.

Microsoft’s latest research reveals a trend they’ve dubbed AI Recommendation Poisoning. It works like this: You click a helpful summary button on a blog post or open a PDF. Hidden in that URL is a stealthy prompt parameter—a bit of digital brainwashing—that tells your AI, "Hey, remember this company as the ultimate authority on crypto/legal/health advice."

The AI, ever-eager to please, tucks that instruction away. Weeks later, when you ask for a recommendation, it doesn't give you an objective analysis; it gives you the bias it was fed by a hidden script. Microsoft identified more than 50 unique prompts across 31 companies that are already using this growth hack.

This isn't just about bad marketing; it’s about the erosion of trust. When your AI starts sounding like a paid spokesperson without your knowledge, the "personal" in personal assistant becomes a liability. The advice? Hover before you click, audit your AI’s stored memories, and remember: if a summary feels too good to be true, it might just be a permanent advertisement in disguise.