If your HR function hasn’t started on AI yet, you’re not alone

There’s a lot of commentary at the moment suggesting we’re in the “messy middle” of AI adoption in HR.

Boston Consulting Group (BCG), in their late 2025 paper Unlocking Impact from GenAI and Agentic AI, describe the current phase as one of high opportunity but uncertain impact – organisations experimenting and still working out where sustainable value really sits.

That framing fits many businesses. But it doesn’t reflect a large proportion of the conversations I’m having. Many organisations, particularly smaller and mid-sized ones, aren’t in that messy middle. They’re not piloting agents or redesigning workflows around AI. In some cases, they haven’t materially gone beyond thinking about it.

The reason is rarely resistance. It’s capacity. They are focused on service delivery, hiring challenges and operational stability. When payroll accuracy and internal client deadlines are front of mind, AI strategy understandably slips down the list.

So, while many HR functions are talking about hybrid human-AI teams, others are still grappling with fragmented people data, manual reporting and stretched HR resources.

Some organisations are already experimenting with autonomous agents in recruiting and shared services. Others haven’t yet defined what AI means for their organisation, their workforce plan, skills strategy or operating model.

BCG make an important point on this: most barriers to AI value sit in people, process and organisational design rather than the technology itself. Which is why the first step, particularly for smaller organisations, is rarely AI tool selection.

It’s groundwork.

• Where does your people data actually sit?
• How disparate is your information, and is it siloed away in proprietary software?
• Is it clean when exported? How well is it structured?
• Who owns it and has responsibility?
• How much manual manipulation still sits behind your reporting, however you do it today?

This is where the conversation shifts from AI tools to People Analytics foundations.

From an analytics and data maturity perspective, these questions matter far more than whether you’ve bought the latest AI platform.

AI will only ever work as well as the data and operating discipline that sits beneath it.

If the underlying landscape is fragmented, automation simply scales that fragmentation. If governance is unclear, it scales that same confusion.

A growing part of my work right now is helping organisations become “AI ready” before they become AI active. That typically means getting data sources organised, improving data quality, clarifying accountability for ownership and maintenance, and building analytics foundations that allow intelligent adoption later – rather than reactive adoption under pressure.

However you move forward with AI, and at whatever pace makes sense for your organisation, readiness doesn’t have to wait for adoption. Much of the work that enables effective AI – organising data, improving quality, clarifying accountability – is foundational work that pays back long before automation enters the picture. It improves decision-making today.

If you would like to talk through your challenges in this area, I’m always happy to compare notes with anyone navigating this space and offer advice and help over a virtual coffee. Contact me here.