BETTER, FASTER, STUCK: WHY AI ISN’T WORKING AND WHAT TO DO NEXT

“Better, faster outcomes is the point. AI just exposes the truth.”

The problem isn’t your AI. It’s your operating model.

Every board is asking about AI. Every budget has an AI line. And yet the core business metrics barely move.

Budgets are approved. Pilots are everywhere. Employees have Copilot and ChatGPT at their fingertips. So why isn’t transformation happening? Research from Harvard Business Review confirms what we see constantly: companies launch AI pilots that save time and make processes more efficient, but those gains don’t scale. Few have been able to fundamentally change their operating and business models around AI. The typical organisation is plugging a cutting-edge AI into a creaky old system, and then wondering why nothing changes.

AI is failing due to organisational limitations. The obstacle isn’t model accuracy or data volume – it’s the murky gap between what the technology can do and how your business actually works day-to-day. Leadership is asking the wrong question. Boards and CEOs press their CIOs: “What’s our AI plan? Are we using enough AI?” The better question is: “How do we need to change the way we work so that AI can deliver real value?” Ask a shallow question, you get a shallow result… like fancy proofs-of-concept that never touch the heart of how the company runs.

Think of it this way: AI is a high-performance engine, but your organisation is the car. If the car has flat tyres and a clogged fuel line, a more powerful engine won’t win you races, it will just spin loudly while the car barely moves. We see companies falling into this trap constantly. They pour millions into AI tools and talent, but nothing about their ways of working changes.

We call this AI theatre – making a show of innovation without addressing the underlying dysfunction. It’s the bank boasting about an AI that pre-fills loan forms, while customers still wait 10 days for approval because five different departments must “check the file.” It’s the retailer using AI to predict hot trends, but failing to get the new stock to store on time because procurement and logistics aren’t aligned. It’s the leadership team proudly announcing a new AI dashboard, yet decisions still crawl through the same four-level hierarchy for sign-off. In each case, AI simply accelerates or amplifies an already broken process.

AI isn’t about making bad decisions faster. Misapplied, it makes a bad situation worse – you get poor results more efficiently.

None of this is to say AI can’t be transformative. It can, but only if you’re willing to transform the organisation alongside it. AI is a catalyst and a mirror, not the hero of the story. It shines a light on what’s not working. If your company is slow, siloed, or plagued by “process debt” – all those accumulated checklists, approvals, and workarounds – AI will expose those issues clearly. And that can be a gift, if you’re ready to act on it. The companies that thrive with AI are using that harsh light to say: “Alright, time to finally fix these bottlenecks.”

Stop putting on AI theatre and start addressing the script, the stage, and the actors. Fix the underlying operating model, how teams are structured, how decisions are made, how work flows, so that when you add AI, the whole machine runs faster, not just one part of it.

At PerfectRebel, we call this the Rewire challenge: changing how your organisation actually delivers, not just how it talks about delivery. It means making your organisation capable of sensing changes and acting on them fast, at scale. Redesigning work processes so that decisions that should take a day don’t take a month. Stripping away the layers of approvals and handoffs that are left over from the 20th-century corporate model. Empowering the people who know the work best to just do the work, supported by AI, rather than pitching it up and down a management chain until it dies.

Crucially, none of this is a one-off “transformation initiative.” It’s a new way of running the company, every day. Many leaders intellectually know this, it’s why so many talk about “breaking silos” and “being agile.” But old habits die hard, and a lot of organisations have simply tried to paper over those silos with technology. That’s the AI theatre we’re warning about.

When new AI meets old ways… and how to break the cycle

A company leaps into AI with a few quick wins,  a team uses generative AI to automate some reports, and productivity in that corner jumps. There’s excitement, maybe a press release about “AI transformation.” But then the reality check hits. To deploy that AI-optimised process at scale, it needs to integrate with five other departments, all with their own priorities, backlogs, and risk checklists. The same old bottlenecks resurface – procurement says the new tool doesn’t fit the vendor policy, the security team raises a flag that leads to an emergency meeting, the legal department insists on a review cycle that’s longer than the AI would ever need.

In our Rebel Perspective podcast, Christoph Burtscher described this exact pattern: the expert team can see a faster path to value, the AI makes it possible, but they still struggle to deliver end-to-end because the organisation around them hasn’t changed. All the usual corporate antibodies come out to slow things down.

The companies pulling ahead are doing two things in tandem: investing in advanced AI and reshaping their operations for speed and flexibility. The litmus test we suggest: stop asking “Are we using AI?” and start asking, “Are we organised to capitalise on AI continuously?” If the honest answer is “not really,” then any big AI initiative is likely to underwhelm.

Consider a common scenario. An engineering group uses an AI coding assistant and suddenly can produce new features in days instead of weeks. But once a feature is ready, it can’t go live until it passes a compliance check, a security review, maybe a change-management board meeting, steps that, in the old way of working, take weeks of emails and meetings. The customer still waits as long as ever.

The efficiency trap: local efficiency, global inefficiency. AI made one slice of work faster, but the end-to-end flow didn’t improve.

Now flip the scenario. What if, from the start, you had redesigned that feature-development process? You put a compliance expert and a security expert in the product team on day one, and you empower that team to deploy to production safely without 20 layers of oversight. Perhaps they even use AI to automate the standard compliance checks in real-time. Suddenly, those two-week delays vanish. A feature idea can go from concept to live in 48 hours, not a month. The difference wasn’t the AI itself, it was changing the way teams are structured and empowered. AI was an enabler, but the real magic was removing the organisational roadblocks.

This is why cross-functional teams are the unit of speed. Working in traditional vertical silos might feel efficient for each department, but as Christoph noted, it quickly grinds to a halt because it’s not cross-functional. In a silo, you optimise for one function’s output, but the moment the task leaves your team, it sits in someone else’s queue. To truly move fast, you have to collapse those queues. That means bringing the necessary people together and giving them the autonomy to execute an entire mission, start to finish. Whether you call it an agile squad, a pod, a tiger team, or just Tuesday’s working group, the label doesn’t matter. What matters is that the team has all the skills it needs to deliver value without constantly waiting on others.

When a cross-functional team is firing on all cylinders with AI at their disposal, it’s incredibly hard for a siloed competitor to keep up. A small, nimble team leveraging AI can experiment, learn, and deliver in a few weeks what takes a traditional structure a year. Don’t benchmark yourself against the usual suspects in your industry, watch out for the small disruptors who are built for agility from the ground up. They will eat your lunch if you stay stuck in slow mode.

A word on governance, because it often comes up: being a fast, AI-enabled organisation does not mean throwing caution to the wind. It means rethinking governance, so it’s baked into the process, rather than a heavy gate at the end. Instead of a separate committee to police everything, you put a compliance or risk expert on the team from the beginning, or you encode the policies into your AI systems, so they flag issues automatically. Embedded guardrails. The team moves fast and stays within bounds, because the checks are part of the engine, not speed bumps on the road.

The organisations that win with AI will be those that change how they work, not just how they tech. As one CEO we know puts it plainly: “AI isn’t going to manage around your org chart.” Exactly. You either redesign the org chart, and the processes and culture around it, to harness AI, or you’ll get minimal benefit from it.

The 90-Day Rebel Test

Suppose you’re a CEO or business leader who’s convinced by all this. You don’t want AI theatre; you want real change. What do you actually do next?

In our Rebel Perspective podcast, we put this directly to Christoph Burtscher, our AI lead, and our co-founder Neil: what is one decision a CEO should make in the next 90 days to avoid wasting money on AI and actually change how the organisation works? Their answer was refreshingly concrete: focus and commit. Pick one meaningful area of your business and go all-in on reinventing it with AI and new ways of working, together.

Why one area? Because trying to transform everything at once usually ends up transforming nothing. Find a high-impact domain,  a product line, a core process, a bottleneck that’s costing you money or customers, and make it your showcase for what better and faster really looks like. Maybe it’s customer onboarding, supply chain planning, or even an internal process like budgeting. The key is that it should be important enough that the outcome matters, and ideally it spans multiple functions, so you’re forced to break some silos as you tackle it.

Next, decide who is going to drive this, because it won’t happen by itself. Build a startup within the company. Pull a few of your best people from different departments, assign them exclusively to this mission, and give them the mandate to challenge the usual rules. Make sure they have an executive sponsor who can clear obstacles and vouch for them at the highest level, because they will ruffle feathers when they start ignoring the old bureaucracy. This isn’t an innovation lab tinkering on the side, it’s a delivery squad tasked with producing a tangible result, fast. They should use AI wherever it’s relevant but also streamline every non-tech step. The goal is to demonstrate, in 90 days, a dramatically faster or better way of doing that piece of the business.

What might that look like? Customer onboarding currently takes four weeks and ten forms. Your strike team redesigns it to one week, mostly digital, with an AI helping to auto-verify documents and a simplified risk-check process co-created with Compliance from the start. In 90 days, they pilot this with a subset of customers and cut the time to 10 days. That’s a significant win. Or if the focus is product development, the team uses AI to generate and test iterations with real customers in a fraction of the usual time. Whatever it is, measure it and show the before-and-after.

Here’s the important part: whether the initiative succeeds or encounters obstacles, you win. If it succeeds, you’ve got a proof point to rally the rest of the organisation, a beacon of what’s possible when you combine AI with new ways of working. If it struggles or fails, treat it as a high-speed learning opportunity. You’ll uncover exactly what policy, what system, or whose buy-in is blocking change. That is incredibly valuable insight to have now, not a year from now.

Either way, you’ve started to shift the conversation from abstract “AI strategy” to concrete operational change. People in your company will take notice. Once one team starts delivering results in weeks that used to take months, everyone else begins to ask: how can we do that too? This is how a new culture spreads, not by decree, but through curiosity. Your job is to double down. Take the lessons and methods from that first squad and start applying them to the next priority, and the next.

Don’t lose nerve if some people grumble that the strike team “broke protocol” or got special treatment. Of course they did. That was the point. If your company’s standard protocol was capable of breakthrough results, you’d already have them. You set up this experiment to shock the system. To show that speed and agility are possible if you’re willing to challenge old assumptions. Celebrate the success, address the valid concerns, and keep pushing forward.

The window for gaining advantage with AI, while competitors flounder, is open right now. Every company is grappling with these issues; many are paralysed or going in circles. Those that take focused, decisive action will separate themselves from the pack. They’ll accumulate learning and benefits faster, and that gap will become very hard to close.

AI can revolutionise your business – but not on its own. It’s waiting for you to revolutionise your business around it.

In the next 90 days, what’s the one bold move you will make to ensure AI actually changes how your organisation works? If you’re ready to stop theorising and start changing how your business actually runs – talk to PerfectRebel.

Source Reference

Harvard Business Review — The “Last Mile” Problem Slowing AI Transformation