The Partnerships AI Will Expose, And the Ones It Will Reward

I have spent most of my career building partnerships. First with corporates and the public sector at Merrill Lynch, UBS and Barclays, then with technology companies at Domotz and Fing. The organizations changed. The job did not.

And every time, I saw the same thing: partnerships that looked strong on paper could still crack once execution began. The cracks came from three places: data, ownership and context. A signed agreement never tests those things.

For years, many of these weaknesses stayed manageable. The systems involved were forgiving of small gaps. AI isn't.

AI Is Exposing Weak Partnerships

I've watched this shift up close. The foundations always mattered - now they show themselves fast.

For years, technology partnerships were judged largely on one question: can our products connect? If data moved between systems and the demo held up, we called it a win.

AI raises that bar. It depends on data that is connected, trusted, and placed in context. It requires clear ownership of that data and a real understanding of the customer environment it describes.

A data gap that might once have caused an operational headache can now affect the quality of an AI recommendation. The same is true of ownership: if an automated decision produces a result nobody can explain or defend, the weakness in the partnership becomes very visible. And then there's context. A context gap surfaces when AI has access to information but not the surrounding understanding needed to make sense of it.

In most organizations, AI is not creating new partnership problems. The partners that benefit most from AI won't necessarily be the ones with the most integrations. They'll be the ones that have got the basics right: trusted data, clear ownership and a good understanding of the customer.

The Data Gap: Why Accuracy Decides Everything

The first gap is data.

KPMG's Q3 2025 AI Pulse Survey found that 82% of executives now name data quality as the top barrier to AI success, up from 56% one quarter earlier.¹ Gartner points in the same direction: organizations with successful AI initiatives invest significantly more in data quality and governance than those with poor outcomes.²

For AI to make an agentic decision - one it takes on its own, not a recommendation a person reviews first - the data has to be correct, complete, and deep enough to carry that judgment alone.

Get that wrong and AI does not just underperform. It acts confidently on the wrong picture. That is worse than not acting at all. Bad data does not just limit an AI capability. It makes it unreliable, because the model can sound equally confident whether the underlying picture is right or wrong.

The Ownership Gap: Why Incentives Have to Point the Same Direction

The second gap is ownership.

Incentives have to point the same direction for years, not just at the signature. A partnership that only pays off when a deal closes behaves nothing like one built for long-term alignment - and that difference is the whole model, not a contract detail. Ours scales with the partner's business, not against it: the more a partner deploys what we build, the more we learn about devices and environments, and the more that learning strengthens the partner's own product over time. In practice, that means building the opportunity together - bringing what we've learned from our own customers straight into the partner's environment, rather than handing them a finished product and a target.

The Context Gap: From Technical Fit to Strategic Fit

The third gap is context.

For most of my career, the first year of any technology partnership went to one question: can we actually integrate our products? Teams spent that time mapping data structures, testing APIs, and discovering, often too late, where the technical assumptions did not hold. The business case waited for the engineers to finish.

AI is changing that. We can now understand much earlier where an integration is likely to work, where the gaps are and roughly what it will take to close them. That means the technical feasibility question is becoming much less of a barrier. Gartner describes this as the arrival of a data feed economy, where verifiable operational data moves between systems as a matter of course and trust, not integration effort, becomes the real gatekeeper.³ It makes the feasibility question fast enough that it stops being the main conversation.

When the technical question gets answered in weeks instead of quarters, the conversation moves somewhere more useful. That's the context gap: whether the combined offering solves a problem the partner's customers actually have, and whether it makes the partner more competitive by delivering real value to the people they serve.

One of our partners builds what they call a "context graph" - combining an organization's content with the context around it to produce insight that people, and increasingly AI agents, can act on. That layer can't come from content alone. It takes understanding the assets and environment underneath it well enough for the insight to hold up - exactly the shift I've watched inside our own partnerships that started out helping a partner populate a configuration database. The product hasn't changed. What it enables has: richer context feeding into that graph, because AI won't just move a process from A to B. It will interpret - and interpretation needs context to land on the right decision.

For me, the best partnerships open doors a partner couldn't open alone - use cases their own roadmap wouldn't reach this year, or next. That's a go-to-market question as much as a product one, and it deserves the same rigor we once reserved for technical due diligence.

Just as important: how does this partnership help the partner navigate the disruption AI is bringing to their own market? And there's another side to this. AI isn't just helping us build partnerships faster. It is changing what our partners' customers expect from them. The partnerships that matter most help a partner stay ahead of that shift, not just distribute a product into a market shifting under their feet.

AI has made "build it ourselves" sound more credible than it used to. Cheaper compute and off-the-shelf models tempt a partner to think they can replicate a capability in-house rather than rely on us for it.

That calculation misses what actually drives the capability forward. Ours doesn't improve only because we invest in it - it improves because it sits inside an expanding network: customers who use it directly, partners who build on it, and their own customers in turn. Every new relationship adds another network, in another industry, with another edge case and new devices - context a single company building alone couldn't accumulate on its own timeline. That advantage doesn't come from hiring more engineers. It compounds because a community keeps contributing to it.

Research from MIT Sloan Management Review makes a related point about AI more broadly. As algorithms, compute, and even talent keep getting commoditized, AI itself is unlikely to be a durable source of advantage once everyone has access to it. What lasts, the authors argue, is residual heterogeneity: the things a company has built up over time that remain out of reach for competitors.⁴ A community of customers and partners that keeps feeding a platform is exactly that kind of asset. So the real question isn't can we build a version of this. It's can we build one with the same breadth of context, starting from zero on years of accumulated contribution. Most of the time, the honest answer is no.

The Question Worth Asking Before You Sign

The partnerships I have seen hold up over years close all three of these gaps at once, not by accident but by design. The data both sides rely on need to be accurate and shared. The ownership of the outcome has to be clear and mutual. And the understanding of the partner's market will evolve as AI changes what their customers expect next.

You won't see any of this in the launch press release. You see it much later - when the partnership is still creating value, or when the two sides have quietly stopped investing in it.

I now ask a simple question early in every partnership conversation: will this make you meaningfully more competitive in your market, not just today, but as your category keeps shifting under AI?

Ultimately, the agreement and the integration are just the starting point. What matters is whether the two companies can actually execute together and create something that neither could have created on its own.

¹ KPMG, AI Quarterly Pulse Survey: Q3 2025, December 2025. kpmg.com

² Gartner, Organizations With Successful AI Initiatives Invest Up to Four Times More in Data and Analytics Foundations, April 2026. gartner.com

³ Gartner, Top Predictions for Data and Analytics in 2026, March 2026. gartner.com

⁴ MIT Sloan Management Review, Why AI Will Not Provide Sustainable Competitive Advantage, David Wingate, Barclay Burns, and Jay B. Barney, 2025. sloanreview.mit.edu