Three blockers are holding up enterprise AI adoption, says OutSystems CEO - and it's not the tech

When I first caught up with Woodson Martin last summer, he was only a few weeks into his new role as CEO of low-code app builder OutSystems. The discussion had the air of a man thinking out loud as he tested ideas for the future of the company, often in conversation with his customers.

When we spoke again as the year drew to a close, I wanted to hear what he’s learned from those customer conversations — and what they reveal about enterprise AI readiness heading into 2026.

The lesson that emerges isn’t that enterprises need persuading about AI — they’ve heard enough vendor evangelism, thought leadership, or exhortations to just ’AI harder’ — but that they need help actually getting it.

Based on Martin’s experience, the biggest issue today isn’t demand, but blockage — a multi-faceted problem encompassing talent gaps, confusion about AI’s purpose, and a noisy solution environment.

Talent is a significant drag on progress

We start with the first of those three blockers, and one of the most persistent that Martin encountered — not machine intelligence but the human kind.

In short, he has discovered that there simply aren’t enough knowledgeable people available to help his customers achieve their AI ambitions. He reveals:

There is a deep hunger for talent, with expertise to be able to do the kinds of things that we're talking about — building applications with a deep understanding of AI.

But as Martin puts it, the problem isn’t just a scarcity of technical skills — it’s a lack of resources able to understand AI’s implications for the company’s operating model:

The necessary resources need strong technical understanding — but also an understanding of the business they’re operating in.

I ask whether talent is currently a hard barrier or just a challenging constraint — and he is keen to point out that customers are still making progress in spite of these challenges:

I think for most it is slowing down, not necessarily a blocker, because they can find resources. But they might be spending more on those resources than they want to — and they can’t find as many as they would like.

Finding some resources doesn’t mean finding the right ones, however, or finding them at the right scale. And elevated costs only further constrain the amount of talent teams can realistically deploy for everything — from sensemaking and narrative creation to architecture and delivery.

Martin points out that a lack of talent therefore means a lack of throughput, no matter how enthusiastic the organization might be:

They can't go as fast as they want. So I'd say for many, it's not like they're in a position where they're completely blocked. It's where they can't go as fast as they want.

But from his explanation, this slow throughput is more damaging than it might first appear — since little visible progress can quickly begin to feel like no progress at all.

What he is describing, then, is not an overly simplistic concern of scarce technical talent — but an end-to-end system of sensemaking, architecture, and delivery that is bunged up across multiple dimensions. And in that that context, even patchy skill gaps don’t just slow delivery — they drain momentum across the organization.

Confusion is a strong source of inertia

The second recurring pattern Martin has found is persistent confusion amongst enterprises about how to apply AI — a situation he says is not helped by the way vendors are communicating:

You know the message we use to communicate, the very basic 10 words we use, are not so different from the 10 words that every other IT vendor is using, right?

But Martin stresses that this is not primarily about feature commoditization among platform vendors specifically — but a much broader blocker rooted in confusion about the true nature of different offerings:

But some of those vendors are using those 10 words to sell services. Some of them are using those same 10 words to sell CRM. Some of them are using those same 10 words to sell ERP.

Martin observes that this leaves customers hesitating to act as they work through their confusion — not only about which is the best vendor but also which is the best kind of vendor:

Speaking in these super broad terms as an industry is creating a lot of confusion. At the end of the day customers say, ‘What’s the difference, and how do we sort through all of this?’

Given that this challenge compounds the lack of skilled resources, I ask Martin how vendors can help address the confusion. His answer is blunt:

Just stop.

Instead, he suggests that vendors should start by stepping out of their own shoes and into those of their customers — helping them to envisage specific and relevant tasks whose outcomes can be significantly improved through the application of AI:

We need to help develop clarity by talking about what customers are actually doing with the tech, what they're accomplishing in actual projects. We need to make things tangible and meaningful.

But Martin also has a second remedy — helping customers understand that most decisions are two-way doors, and that you don’t need to have resolved all confusion before you start to make progress:

We need to really communicate a strong sense of momentum, that this is something you can go do now without actually figuring out all the things just to get started.

Internal traction is a growing challenge for CIOs

The third blocker Martin describes falls onto the shoulders of CIOs — whose teams are often ill-equipped to respond.

He illustrates the problem by sharing the case of one CIO in the Netherlands:

He called and said ‘Help me sell the platform we’ve built’. Every time they tried to explain the platform, a busload of people from Accenture or Deloitte would turn up, hawking Salesforce or ServiceNow as an alternative.

And Martin suggests this isn’t just a question of scale, but of narrative skill:

He told me that he’s trying to compete by sending a program manager to meet with the team — but that vendors show up with a busload of consultants. But that tension is also natural — business owners ask everyone for the same things. I mean you have to be loud in this industry.

The resulting narrative vacuum allows consultancies, vendors, and platforms to crowd in with credible but incompatible answers — creating a noisy solution environment where business leaders face too many plausible paths and no clear way to choose.

But the language CIOs use when asking for communication support is often revealing, according to Martin:

And so they're like, help me evangelize OutSystems as an agent platform — and that Workbench is a technology to really get traction with an agentic system.

He suggests this kind of technical language does little to help time-poor business leaders commit to a direction — instead eroding confidence and sending them in search of simpler explanations elsewhere.

And while Martin says the approach is flawed, he applauds the focus on mindshare, which he argues is critical to taking control of the transformation narrative and removing barriers to progress — but which he suggests needs examples, not technology:

The start of the demo shouldn’t be the developer persona. The start of the demo needs to be a solution that makes the P&L owner go, ’Wow, I could dramatically improve the productivity of my organization.’ Because there's a lot of noise and you’ve got to cut through it by setting a vision that creates clarity for the platform strategy.

This third form of blockage comes down to too many plausible solution options creating organizational paralysis.

My take

Taken together, Martin’s observations give a live field perspective on where companies are with AI adoption — and the story is both more positive and less positive than a simple reading of market sentiment would suggest.

More positive because the appetite to act is there. Less positive because the capability to act is not.

But at least — while painful — capability development is just a matter of commitment and graft. Given the challenges Martin’s customers are facing in absorbing AI, perhaps the focus of this graft should be more on the adoption and less on the AI.

Because the challenges he has encountered — talent, confusion, and overwhelm — are systems challenges, and therefore do not come alone. Instead, each compounds the others — creating a kind of ‘anti-flywheel’ that diminishes certainty, momentum, and energy with each painful turn of the wheel:

• Without the right talent companies cannot understand what to implement, sell the transformation narrative, or deliver.
• Without certainty of purpose companies cannot move decisively, acquire the necessary talent, or have a clear context for transformation narratives.
• Without a clear transformation narrative, companies cannot choose a path, steer talent strategies, or escape the paralysis of overwhelming choice.

And these blockers matter — not just for the companies involved but for the wider industry and economy.

Enormous sums are now being invested by vendors, platforms, and infrastructure providers in the expectation that capacity will translate smoothly into adoption and value — an expectation that isn’t currently being met according to a range of industry reports.

But Martin’s fieldwork suggests that the reported lack of AI progress is not because the technologies are weak, but because the translation layer between raw potential and polished business outcome is breaking down — a more subtle and realistic explanation for why progress is slower than many expected, perhaps.

More generally, however, his experience surfaces three signals that organizations should be aware of if they are to unlock this problematic translation layer.

Firstly, companies need talent that doesn’t only understand technology. They also need people that understand their business context. While buying in resources might give a quick hit, it is difficult, expensive, and — most critically — doesn’t address this core challenge of business context. AI is first and foremost an operating model transformation — and so business context needs to precede technology rather than follow it. And while vendor-led professional services may help at the margins, they don’t remove the underlying dependency on business expertise. It is therefore likely that the most sustainable responses to talent scarcity will involve a much stronger focus on developing capability from within.

Secondly, organizations need to become far more discriminating around use cases. In particular Martin’s experience suggests that they should ignore vendor claims unless they are couched in outcome-focused language. There should be no AI programs — because AI is a tool, not a purpose. Instead, organizations need to anchor decisions in outcomes first, and treat AI as one of several possible means. In this context, Martin’s commitment to ’just stop’ restating the same 10 generic AI claims as every other vendor is welcome — a pivot that was evident at the company’s event in Miami.

Thirdly, CIOs need to become narrative players. It is no longer sufficient to wait for the business to ask, or to assume that — as the internal technology department — you have a monopoly on delivery. CIOs and their teams increasingly need to think about narrative, persuasion, and positioning — not just delivery. Because technology is now digital business capability — and there are plenty of external organizations ready to expend significant financial and relational capital to win over business owners with expansive (and expensive) narratives. The growing calls for Martin to help with this process suggest that CIOs are also expecting more help with internal selling from their chosen vendors — not via the deployment of platforms, but via the deployment of credible stories that help those platforms gain internal traction.

Stepping back, therefore, Martin’s conversations suggest to me that enterprise AI has reached an awkward adolescent phase.

Organizations know what they want, but lack the organizational maturity to get it.

Not because they don’t want AI — but because, as adolescence teaches us, simply wanting something isn’t enough to get it.