Eighty-three percent of operations leaders now believe AI and automation will accelerate the breakdown of the silos that have shaped how their organisations work for decades, according to PwC’s 2026 Digital Trends in Operations survey.
It’s a confident number. Then comes the number underneath it: only 41% of those same organisations say they actually operate with a collaborative, horizontal structure today. And of the specific combination that matters most, AI genuinely embedded enterprise-wide, with no real barriers to scaling it, a horizontal operating structure, and technology investment delivering what was promised, only 4% of firms get there.
That gap between belief and reality isn’t a technology problem. It’s what happens when everyone buys the tools and nobody actually agrees how the teams around them are supposed to work together.
We saw this up close at our own AI Capability Building session in Belfast this summer. One participant described running an internal AI amnesty, simply asking staff what tools they were actually using, which versions, whether personal or corporate. What came back was a landscape of developers on one model, product managers on another, QA teams pulling data into tools that had never been through any formal approval. Nobody had planned it that way. It had just accumulated, team by team, decision by decision, faster than anyone was tracking it.
The instinct, when that picture emerges, is to centralise everything and lock it down. The room pushed back hard on that instinct, for a reason worth sitting with: tell a fast-moving team you’re taking away the tools they’ve already built workflows around, and you don’t get compliance – you get frustration and workarounds. The more useful answer wasn’t prohibition. It was a governed default: agree the tools available across the organisation, apply role-appropriate access, and make experimentation outside that sanctioned set clearly out of policy – a shared, understood boundary rather than a locked door nobody explains.
That’s a collaboration problem before it’s a technology problem. It only works if the teams affected actually agree on where the boundary sits and why, which is precisely the “shared purpose” that so many cross-team technology initiatives skip past at kick-off and never revisit.
Separately, a 2026 survey of 150 financial services firms found 61% say AI has fallen short of the ROI they expected for the cost and effort invested, despite the sector leading the market on formal AI governance frameworks and adoption pace.
The firms getting real value weren’t the ones with the most advanced tooling. They were the ones with dedicated capacity for the unglamorous work after launch: monitoring, managing the data flowing in and out, refining as conditions shift, work that only happens when the teams responsible for it are genuinely coordinated, not each defending their own patch of the initiative.
This is the pattern underneath most stalled cross-team technology work: success gets defined loosely at the start, each party assumes the others understand their role, and nobody revisits any of it once the project is underway. Friction shows up eventually as delay, a missed handover, a decision nobody remembers agreeing to, a partner team optimising for a different definition of “done.” By the time it’s visible, it’s already cost weeks.
The organisations getting ahead of this aren’t relying on better tooling alone. They’re treating shared purpose as something to define explicitly with every party at the table, not just the leads, and revisiting it at real milestones rather than assuming the kick-off conversation still holds six months later.
And critically, they’re treating cultural fit and working-style alignment across collaborating teams and partners as something to actually check, rather than something everyone politely assumes is fine until it isn’t.
None of that requires better AI. It requires the same discipline PwC’s data points to: connecting technology across segments and teams as a deliberate structural choice, not a byproduct of everyone independently buying good tools.
Our free 15-minute Leading through Disruption diagnostic, developed in partnership with SolasOS, includes a dedicated section on cross-team collaboration and shared purpose, an instant, evidence-based read on where your technology initiatives are exposed.
Start the diagnostic: solas-os.ai/readiness
Next in this series: why most AI workforce transformation plans are built on job descriptions instead of where the work actually happens.
This article draws on discussion from FinTech-Tables’ AI Capability Building session, AI & Data Leadership Forum (Belfast, June 2026), alongside research from PwC’s 2026 Digital Trends in Operations survey and Coastal’s 2026 AI Operations Report: Financial Services.
