Laurence CadieuxAlmost everyone is using AI now. In our upcoming 2026 AI Maturity Benchmark Survey, only 8% of organizations told us they aren’t using AI at all. For the other 92%, it’s already part of the daily routine: content creation, software development, data analysis, IT operations, security, documentation, support.
But here’s the thing we kept coming back to as we dug into the data: using AI and being good at using AI are two very different stories.
An organization can pile up tools without a strategy, without the right skills, without real integration, and without any way to know whether those tools are actually paying off. So instead of asking who uses AI, we asked a harder question: what actually changes as organizations get more mature with it?
We surveyed 616 IT professionals across different roles, company sizes, industries, and regions, then grouped them by self-reported maturity:
- Early-stage
- AI Adopters
- Advanced
The full report drops in September. Until then, here are four findings we think are worth sitting with.
A quick word on timing (and a thank-you)
We opened the survey on July 7 and closed it a couple of weeks ago: a solid month of data collection. We’re turning the report around fast on purpose. AI is moving quickly, and a benchmark is most useful while it still reflects where things actually stand.
Thank you to everyone who answered. 616 responses is a strong sample, and your open-ended answers especially helped turn percentages into a real picture.
If you took our AI survey when it was open, this sneak peek is for you.
Leadership support isn’t a nice-to-have; it’s the dividing line
You can start using AI without leadership. Someone tries a tool, a developer wires in an assistant, an IT team finds a use case. But getting past scattered experiments takes something else: direction from the top.
The gap here is bigger than we expected. Only 14% of Early-stage organizations say leadership actively promotes and invests in AI. Among Advanced organizations, that number is 91%: roughly 6.5 times higher.
The interesting part isn’t just that support grows. It’s what kind of support people ask for. Early on, teams struggle with confusion about where to even start. In the middle, leadership is enthusiastic but the destination is fuzzy. By the Advanced stage, support gets concrete: leaders help people learn, experiment, and build real confidence. The progression isn’t “leadership suddenly has all the answers”; it’s support becoming clearer and more useful.
That tracks with what we’ve shared about our own approach: AI is encouraged on purpose, with leadership backing and structure behind the tools.
Scaling AI isn’t about doing more; it’s about spreading wider
As organizations mature, they don’t necessarily discover brand-new categories of AI use. They extend AI across more of their work at the same time.
On average, Early-stage organizations use AI in about 2 of the 9 areas we tracked. That climbs to nearly 4 among AI Adopters and more than 6 among Advanced organizations. Content creation alone jumps from 47.6% of Early-stage organizations to 83% of Advanced ones.
And that breadth comes with a catch. The more places AI shows up, the harder it gets to connect everything securely and consistently. As one Advanced respondent put it, the challenge isn’t the AI models themselves; it’s integrating them into a secure, unified platform instead of juggling a pile of separate tools and custom connections. Scaling AI doesn’t remove the hard parts. It relocates them.
AI agents are closer than you might think
Agents (AI that can actually take action across tasks and workflows, not just generate or summarize) are moving from “someday” to “already happening.”
The split by maturity is striking. Among Early-stage organizations, about 7 in 10 haven’t moved past exploration. But among Advanced organizations, more than 4 in 5 are already piloting agents or running them in production. The picture almost completely flips as maturity rises.
That said, maturity doesn’t erase caution. Security stays the top concern across every group, and loss of human control ranks second overall, even among the most advanced organizations. What people describe isn’t humans disappearing from the process. It’s the human role shifting toward judgment, context, review, and accountability. As one respondent put it: the most valuable IT professionals won’t compete with AI on speed; they’ll use AI while owning the final decision.
The takeaway (and what’s coming)
Across all four findings, one pattern holds: maturity isn’t about having more AI. It’s about how deliberately an organization leads it, owns it, scales it, and measures it.
One thing worth repeating: this report shows what travels together with higher maturity, not what causes it. Association isn’t causation. But the patterns are consistent enough to be genuinely useful if you’re trying to figure out your own next step.
This is just a preview. The full 2026 AI Maturity Benchmark Survey, with all five insights, the open-ended responses behind the numbers, and practical recommendations, lands in September.
Want to be first to read it? Keep an eye on the Devolutions blog, or sign up for our newsletter so the report hits your inbox the day it’s out.
Questions or reactions to these early findings? Leave a comment below, or join the conversation on the Devolutions Forum.
Steven Lafortune
