AI Made People Faster, Not Teams Better: 69% of AI Work Happens in Private

Ask almost anyone at work whether AI has made them faster, and they will say yes. Ask whether it has made their team better, and the answer gets a lot less certain.

Miro's new State of Intelligent Collaboration study, published November 3, 2026, measures that gap. Miro surveyed more than 9,500 leaders and employees across North America, Latin America, Europe and Asia-Pacific in September and October 2026, from product development, IT, operations and other functions. As a collaboration software vendor, Miro has an interest in the answer, so it is worth reading the numbers for what they show rather than for the product pitch.

What they show is clear: AI adoption scaled one person at a time, and the work stayed that way.

Individuals Got Faster. Quality and Teamwork Lagged.

From Miro's summary of the findings:

• 83% agree AI has made individuals more productive.
• 73% say their speed of output has increased.
• Only 56% say quality has improved, and most have yet to see better decision-making or collaboration.

Individual AI use is also still shallow. 64% of knowledge workers use AI mainly for information or task help, like asking questions or getting first drafts. Only 17% have reached partly or fully autonomous workflows.

69% of AI Work Happens in Private

The study's most telling number is about where AI work happens. 69% of respondents say most AI-assisted work happens in private sessions rather than shared spaces.

That makes sense given how most people started using AI: each person with their own tools, their own chat history and their own way of prompting. But it has costs that show up at the team level. When the context, the drafts and the reasoning live in one person's chat window:

• Nobody else can reuse a prompt or process that works.
• Teammates review the output without seeing how it was made.
• The same question gets asked, and answered differently, by five people.
• Leaders cannot see what is being used or how.

That last point shows up in the data. 40% of leaders rate their visibility into AI adoption, usage and best practices as poor or non-existent.

The Hidden Costs Land on Teams

Miro's respondents describe what fragmented AI use feels like day to day:

• 59% report more AI-generated output to review.
• 55% report more "AI slop," low-quality output that someone has to catch.
• 44% say context switching has increased in the past 12 months.
• 60% say AI has blurred previously distinct roles, and 44% say company culture has become less collaborative.

At the leadership level, half struggle with information and functional silos. 47% call out slow decision-making, 45% point to longer review and feedback cycles, and 44% have seen decisions made without all the relevant context. Between 53% and 58% flag privacy, compliance, security and governance as top challenges.

Organizations Are Further Ahead Than Their People

On paper, companies are moving. A quarter of respondents say AI is scaled across functions or has transformed their operating model, and over a third say it is embedded in select processes with defined ownership. Only 16% are still experimenting ad hoc.

But Miro found a pattern across regions and industries: "the further ahead an organization is with AI implementation, the further behind its people tend to be." Rolling out licenses is the easy part. Getting a team to work with AI together is the hard part.

Sign up for our newsletter

Get the latest updates, product launches, and helpful tips delivered straight to your inbox.

Email addressSubscribe

How to Make AI a Team Sport

Miro's own top recommendation is to "make shared workflows the unit of AI adoption, rather than individual tool preferences." In practice, that looks like:

• Put AI where the team already works. An agent that posts its work in a shared Slack or Teams channel is visible by default. A private chat window is not.
• Give agents shared context. Connect agents to the team's actual documents, tickets and records, so everyone gets answers from the same source instead of from whatever each person pasted in.
• Turn good prompts into shared routines. When one person figures out a process that works, make it a standing routine the whole team can use.
• Make review part of the flow. If AI output is landing on someone's desk to check, decide who checks what, and keep that step visible.
• Build visibility in from day one. Clear ownership and admin controls make it possible to see which agents exist, what they can reach, and what they do.

Where Workmate Fits

This is the problem Workmate is built around. Instead of every person running their own private AI sessions, Workmate builds custom AI agents for a team's real workflows, connects them to the apps and data the team already uses, and launches them inside Slack, Microsoft Teams and email. The agents work in the same channels people do, so their output is visible to the whole team instead of sitting in one person's chat history.

Admins control access, app connections and skill installations for every agent, which addresses the visibility gap Miro's leaders describe. Credentials are stored in a vault that can require human approval, and Workmate is SOC 2 Type II audited.

What This Means Going Forward

The first wave of workplace AI made individuals faster. Miro's data suggests that wave has mostly run its course: more speed alone is now producing more output to review, more slop to catch, and more context switching.

The next gains will come from teams that move AI out of private chats and into shared workflows, where the context, the work and the review are visible to everyone who depends on them.

Frequently Asked Questions

What did Miro's State of Intelligent Collaboration study find?

Miro surveyed 9,500+ leaders and employees in September and October 2026. It found 83% say AI made individuals more productive, but only 56% say quality improved, and 69% say most AI-assisted work happens in private sessions.

Why isn't AI improving team performance?

According to Miro's study, adoption scaled person by person, with separate tools and chat histories. That leaves teams reviewing more output (59%) and more low-quality "AI slop" (55%) without shared context.

How many workers use AI agents or autonomous workflows?

Only 17% of knowledge workers in Miro's study have reached partly or fully autonomous AI workflows. 64% use AI mainly for questions and first drafts.

How can teams collaborate better with AI?

Run AI in shared spaces the team already uses, connect agents to shared data, turn working prompts into team routines, and build in clear review steps and admin visibility.