What changed, what worked, and where leaders are heading next

In 2025, AI stopped being something teams were “trying out” and became something leaders had to take responsibility for. 

This recap looks at what changed over the last year, what delivered real results in practice, and what organisations need in place to scale AI sensibly in 2026. 

From headlines to hard questions 

If 2024 was full of AI hype, 2025 was the year leaders started asking tougher questions. 

Not: “Should we be using AI?” 
But: 
“Where is it actually helping?” 
“What value are we getting back?” 
And why hasn’t more of it stuck yet?” 

Across our work with retail, housing and service organisations, one thing became clear – AI is no longer a side experiment. But it only works when the basics are in place. 

What changed in 2025 

From proof of concept to proof it works 

This year, we saw organisations move beyond pilots and demos and start building for the long term. A few big shifts stood out. 

1. AI moved out of innovation teams and into leadership conversations 

In housing and care especially, organisations moved from curiosity to clear direction. 

AI stopped being treated as a shiny tool and started being treated as a capability – something tied to real priorities like improving tenant experience, managing assets better, and meeting long-term sustainability goals. 

Several organisations have now committed to new AI initiatives for 2026, alongside clearer rules, ownership, and plans for how AI fits into their wider roadmap. That shift, from “let’s try this” to “let’s do this properly”, is what will separate leaders from laggards next year. 

We saw this most clearly in housing and care, where the conversation moved from “can we do this?” to “how do we roll this out safely, consistently, and without disrupting frontline teams?” 

2. People and process mattered more than the tech 

The biggest blockers we heard about this year weren’t technical. 

They were things like: 

• Not enough internal skills to keep AI running and improving 

• No clear owners for data or decisions 

• Systems that didn’t talk to each other, slowing everything down 

From retail roundtables to delivery workshops, the message was consistent: getting the people, roles and basics right matters far more than choosing the “right” tool. 

3. AI is now expected to earn its place 

Working demos aren’t enough anymore. Leaders want impact. 

We saw far more focus on: 

• How value will actually be measured 

• Whether results can be repeated, not just one-offs 

• How AI supports wider business change, not just isolated wins 

The organisations seeing the best outcomes didn’t start with technology. They started with a clear business problem and worked backwards from there. 

What worked in practice 

Automation on the frontline: job reconciliation 

One client automated a manual job review process that had been slowing teams down. 

The results were clear: 

• Data collection moved from 10 minutes to less than 2 minutes 

• Manual review and what used to take more than 20 minutes now takes about 2-3 minutes 

• Weekly capacity that fulfilled 250 jobs can now fulfill 5,000 jobs 

That’s a 20x increase in productivity and it’s now shaping how similar processes are tackled elsewhere in the organisation. 

Talking to your data (without waiting on reports) 

A large retailer wanted teams to stop relying on analysts just to answer everyday questions. 

Instead, teams can now ask things like: 

“What were sales by region last month?” …and get a clear, visual answer in seconds with confidence in where the numbers come from. 

Teams can get answers themselves, without waiting on dashboards to be updated or relying on data teams to step in. 

Making AI fit into retail, not disrupt it 

In complex retail environments, success came from fitting AI into existing ways of working; not bolting on new tools. 

The focus was on helping teams move faster using the systems they already rely on, rather than asking them to learn entirely new workflows. Additionally, branch teams were involved early in the work, helping shape the AI so its outputs matched how decisions are actually made day-to-day. 

When AI supports day-to-day decisions instead of sitting on the sidelines, adoption happens naturally. 

Faster creative without losing brand control 

More retail teams are using AI to create product imagery in consistent, brand-safe settings. 

That’s cutting down studio time, speeding up campaigns, and giving teams more flexibility without compromising quality or control. 

Supporting frontline care teams 

In large care organisations, AI was used to reduce manual effort and support decision-making in fast-paced environments. 

Instead of experimenting in isolation, use cases were designed around how frontline teams actually work. Bringing those teams in early helped build trust, shape the right solutions, and speed up adoption, which matters hugely in settings where confidence and consistency are critical. 

What didn’t work (and what we learned) 

No year in AI is complete without a few hard lessons. 

At one client, delays in accessing data led to project overruns. The takeaway here is that if the data isn’t ready, the project won’t be either. 

Recommended Reading: How to Get Your Data AI-Ready: Practical Tips for Better Forecasting, Happier Customers, and Smoother Operations

Across multiple projects, we saw the same issue repeat – unclear data ownership and access slows everything down. Getting this agreed upfront saves time, money, and frustration later. 

In another organisation, progress stalled because key data was owned by third-party vendors. The teams were ready, but their hands were tied. The lesson? If you don’t control your data, you don’t control the outcome. This needs to be part of vendor decisions from day one. 

Frontline engagement mattered more than expected 

In care settings especially, even well-designed solutions struggled when teams weren’t involved early enough. Adoption slowed, and trust took longer to build. 

The message is clear: bringing frontline teams along isn’t a “nice to have” – it’s essential. 

Connecting systems is harder than it looks 

In large retail organisations, linking AI across multiple systems proved more complex than expected. Disconnected data and older processes added friction and slowed progress. 

Time and again, we saw the same lesson: planning how everything fits together upfront beats trying to fix it later. 

What leaders are focusing on in 2026 

1. Scaling what already works 

Instead of chasing new ideas, leaders are asking: 
“What should we expand?” 

That means building on proven use cases, assigning clear owners, and putting the right foundations in place to support growth. 

2. Readiness will separate the winners 

It’s no longer about who adopts AI first. It’s about who’s ready to use it properly. 

That includes: 

• Data teams can trust 

• People who understand how to use AI day-to-day 

• Systems that work together, not in silos 

Recommended Reading: How Retailers Become Data Confident – A Simple, Practical Guide to Getting Your Data Ready to “Talk” Back

3. People will define success 

The tools are improving fast. The real questions now are human ones: 
Who owns this? 
Who understands it? 
And can teams trust the outputs? 

In both retail and consumer services, 2026 will be shaped by how well organisations build AI into everyday operations; not how impressive the technology looks. 

Final thought 

2025 showed that AI isn’t about buzzwords anymore. 

It’s about doing the hard, unglamorous work – embedding AI into decisions, processes, and teams safely and with purpose. 

The organisations seeing real returns aren’t chasing headlines. They’re building something that lasts. 

Planning your 2026 roadmap? We’re already helping leadership teams prioritise the right opportunities and set themselves up for long-term success.