Why the clock is ticking to get your AI strategy in place

When 2026 began the industry took a collective breath. An AI onslaught overwhelmed. Vendors, platforms and tools flooded the zone. We knew there was good stuff in there – but we needed to figure out what.

So, agencies and brands agreed to step back for 12 months. That tacit understanding gave us breathing space to experiment and to learn. No one I know calls this grace period unwelcome. We’re collectively using it to discover what works and what doesn’t. Even – contrary to agencies’ age-old rules of engagement – to share ideas.

But time is ticking by. We’re almost halfway through the year. Soon thoughts will turn to 2027 spend and strategy. Expectations will ramp up, budget calls will be made. If your AI plan isn’t clear by Christmas, you’ll get left behind.

For AI strategy the next six months are even more crucial than those already behind us.

Learnings from AI application
The “Year of AI Learning” has so far been collaborative, inspirational – and fun. It’s also worth saying that at no point have our people felt under pressure from the threat of AI stealing their jobs.

If anything, the tech makes human input more important. It’s true that repetitive processes (document iteration, forecasting, mass emails to name a few) can and are being automated. But the outputs require a new level of quality assurance.

With a lean but strong team in place, and feeling secure, we began to experiment with all flavours of AI: generative, agentic and analytical. We did this on behalf of our clients and for internal purposes, too.

After hundreds of hours spent exploring thousands of tools here are some of the things we’ve learned:

Never operate alone. It might sound obvious, but when there’s a shiny new toy in town it’s easy to become isolated in thoughts and activities around it. A shared vision and collegiate effort to find solutions, involving all relevant agency and client teams, is critical to find a mutually beneficial AI strategy which can take both businesses forward.

Don’t overpromise. Closely related to the previous point is the tendency for agencies to default to “we can do everything” in a bid to impress or reassure. It isn’t possible to possess all of AI’s answers at this stage. Much better to collaborate with the client, uncover their unique challenges and design possible solutions accordingly.

Pick the right projects. The possibilities of AI are almost endless, so where do you begin? The answer is what it has always been. Start with a clear and deep understanding of client challenges and goals. Align AI’s capabilities with them to create a task list. Then pilot, pilot, pilot. Spin up projects that won’t always succeed, but can run simultaneously and quickly point the way to what should be developed.

Just do it and see what sticks. At this stage it’s fine to give yourself permission to test, tinker and try as many new approaches as possible – whether with agentic, generative or analytical platforms. Some of the best discoveries are made through experimentation alone. While time is still on our side we should embrace the widest possible scope for AI pilots. The tech is a great leveller: what someone new to the industry uncovers can be just as valuable as the work of a leader with decades of experience.

Only ‘tools for all’ will work. If investment in AI is to pay off, tools the tech enables must be accessible to as many parts of your organisation as possible. Think of Slack. The amusing part of its success is the fact it’s a project management tool! But its ease of use and productivity boost have attached emotional energy to the platform. The more people you make AI tools for, the more successful those innovations will become.

Fee structures are changing fast. Activity levels which software credits secured for your team last summer were already shrinking by the new year (as some of us predicted they would). This means AI efficiencies now require more budget after all. Human hours aren’t as undercut as expected – a balance will continue to be vital.

There’s plenty to think about as the year of AI learning reaches half time. And I haven’t even covered what comes next: most likely GEO strategy, where taking back control from LLMs over how and when content is surfaced is a growing consideration. But the good news is, time isn’t up. We’re all making huge progress though the new AI-shaped landscape. We just need to move even faster in the six months ahead.

Charlie Bruton is director of innovation, AI and technology at Iris London