Last month, I spent a day at FutureWeek Forum in London, centred around the future of AI, advertising, and commerce. The conference brought together some of the sharpest minds across AI, advertising, and commerce, and the conversations were refreshingly nuanced. The signal was clear: AI is helping the industry evolve quickly, but the fundamentals haven’t changed. Humans remain at the heart of creating value and meaning in an increasingly complex space.

Across sessions, disciplines, and even agency-brand boundaries, a few themes consistently surfaced. Here are the main things I took away about how the industry can work, plan, and grow.

Trust is the new currency

If there was one word that defined the day, it was trust. Not in the abstract, mission-statement sense, but in the practical sense of when AI feels helpful and when it feels intrusive. Given the understandable concern surrounding AI, that line is often extremely thin.

An AI assistant that knows your schedule well enough to pre-book travel feels useful, but an AI bot that gets something wrong, or more worryingly, too right, can permanently damage trust. Personalisation only works when it feels genuinely helpful, and that will increasingly become the difference between adoption and backlash as agentic systems, synthetic audiences, and black-box buying become more common.

That tension also needs to remain front of mind when building guardrails around these systems, particularly as automation becomes more embedded in decision-making and discovery.

Brands still win the shortlist

There is a growing narrative in the industry that precision targeting makes brand-building obsolete. Why invest in awareness when an algorithm can find the right person at the right moment and drive conversion? I’ve always found that claim dubious, and in the context of LLM-driven search results, it feels even further from the point.

Think about the user experience. A user comes to an AI platform with a problem. The LLM shortlists products that could solve it. But ultimately, deep-rooted perceptions and biases still drive the final click and decision. In an LLM-mediated discovery world, brand equity becomes a brand’s best defence.

That means awareness spend isn’t just still relevant, it may now be the entry fee for consideration. For challenger brands, the stakes become significantly higher.

“GEO is just good SEO”

This was possibly my favourite line of the day, from Xanthe Pickford-Avery, which she covered during the panel on Brand Discovery in Chatbots, with Mastercard’s Lidewij van den Ham. The fundamentals haven’t changed, but the language has.

When people use LLMs to discover products, they don’t search the same way they do on traditional search engines. Users search for outcomes rather than products, which means Generative Engine Optimisation strategies need to evolve accordingly. But the core principle remains the same: understanding intent, context, and the language people naturally use.

For LLMs to surface and trust your content, it needs to be answer-first and focused on solving problems rather than leading with products. This isn’t a passing trend; it’s a structural shift in how discovery works.

Agencies' real value is making sense of chaos

One of the more interesting discussions came from the panel on Reimagining the Agency Value Proposition in the Age of AI. The conclusion wasn’t that agencies are under threat, but that their value is becoming clearer.

As AI automates more elements of planning and trading, agencies become more valuable for their ability to navigate complexity, apply human judgement where AI falls short, and translate fragmented signals into clear strategic direction.

That means making sense of disconnected datasets, channel proliferation, and emerging protocols like ACP and AAO, while helping clients prioritise what matters. The role of the agency becomes less about execution alone and more about the integration of interpreting nuance, navigating ambiguity, and identifying the signals that matter most when the black box falls short.

That isn’t a diminished role. If anything, it’s a more strategic one.

AI augments shopping, it doesn't replace it

During the discussion around Agentic Commerce and the Zero-Click Customer journey, panellists mentioned Agentic commerce works best when it delivers convenience: planning multi-stop trips, comparing insurance options, or streamlining routine decisions; but it is unlikely to remove the emotional side of shopping or brand engagement.

People will still want to browse when the experience itself brings value or enjoyment, whether that’s shopping for a wedding guest dress or discovering something they didn’t know they wanted. Brand experiences still matter because discovery is not always purely transactional.

AI-driven commerce is a new channel with different rules, but it doesn’t replace every other path to purchase or make broader advertising irrelevant. Marketers can still create moments of discovery and emotional connection that drive impact.

So, what does it mean?

Across Programmatic, Planning, Commerce, and Social, the opportunity is the same: use AI to do more but lead with strategy. Clients don’t just need people who can operate the tools. They need partners who understand which tools to use, when to use them, why, and to safeguard their brand.

This also feels like the moment to double down on brand voice and creative integrity. In an AI-saturated landscape, where content risks becoming increasingly interchangeable, distinctiveness becomes one of the few lasting competitive advantages brands can build.

AI is expanding the toolkit, not replacing strategy or human judgement. Automation can accelerate execution and scale output, but the quality of the work will still depend on the thinking, creativity, and intent behind it. Brand building, emotional connection, and strategic clarity remain central.

The agencies and teams that thrive will be the ones using AI to automate transactional work so they can focus more deeply on the strategic and human parts of the customer journey that machines still can’t replicate.