Is your 2026 AI strategy focused on efficiency? Wrong.

Next week, we celebrate three years of AI going mainstream. The journey was distinct: Year 1 was hand-raisers writing impact papers; Year 2 saw firms seduced by Microsoft’s sales machine into adopting Copilot; and this year brought disillusionment after buying ChatGPT seats that haven't yet transformed the business.

I bet you are in the middle of strategy offsites where AI is back on the agenda.

We see a market stuck in random experimentation. Having assessed over 50 advisory firms, we learned that while AI is seen as important, it isn't treated as urgent. Partners have great ambitions but are too busy doing deals. The challenge is moving from random experiments to a structured approach.

If your 2026 plan focuses on efficiency—writing emails or summarizing slides—you are aiming too low. Efficiency is a race to the bottom; eventually, all firms will get efficient, and the advantage erodes. To set yourself apart, focus on value, not cost.

My two cents for your strategy:

1. Focus on value 

We picked the low-hanging productivity fruit; now we must go where others can't. If your client hires you to sell their company for a higher price, AI should assist there—enhancing expertise, not just lowering costs. Accenture and McKinsey agree: high performers set innovation goals, whereas average performers get stuck on efficiency.

2. Build AI literacy and a culture of experimentation Don't train everyone at once. Identify your top 10% power-users who are naturally curious. Give them time and tools to experiment and find value, then let them coach the rest of the organization. This is about co-creating new ways of working and reinvention from the inside out.

3. An AI stack enabling human-agent collaboration 

You need more than a chat interface; you need a robust digital core. Build a single platform where human experts and AI agents truly collaborate. This allows you to orchestrate the best tools and LLMs for the task. It isn't just about data—it is the 'last mile' of bringing data to life in a secure environment where people and agents work side-by-side.

A final thought: Don't get me wrong, reducing costs is welcome. But when you implement AI to solve high-value problems, efficiency happens as a natural by-product. You get speed for free; plan for value.