A survey of 798 employees at Russian companies
Work is moving to personal accounts: shadow AI in Russian companies
88% of employees already use AI, yet only one in four says their company has officially rolled it out. So employees are taking matters into their own hands: more than half of them use AI through personal accounts. We surveyed 798 people at Russian companies, from engineering and sales to HR, training and finance, to see how AI actually fits into their work.
This study was conducted by the team behind Pachca, a team chat app where employees and AI agents work side by side in one workspace, inside your company's infrastructure. Building your own agents? Check out our developer guide.
Five numbers that sum it up
Based on responses from 798 employees at 100+ Russian companies.
01AI is already at work
Half of employees use AI every day
88% use AI at least occasionally. But here's what matters more: 69% use it daily or several times a week. For them, it's no longer a toy but a go-to work tool.
All respondents: 798 people
02Companies are falling behind
Only one in four employees says their company has officially rolled out AI
Employees already work with AI, and companies can't keep up. Most say nothing has been rolled out, or that “someone on another team is trying something.”
say it's officially rolled out, with tools and policies
The other 76% said AI is only partly rolled out, not rolled out at all or that they don't know:
AI users: about 708 people
Definition
Shadow AI is what this study calls using a personal AI account for work tasks.
03Work slips into the shadows
58% of AI users rely on shadow AI
36% of AI users regularly handle work tasks through personal accounts, and another 22% do so from time to time.
AI users: about 708 people
04Which models people choose
Personal accounts lean toward foreign models
Among those who regularly use a personal account, 88% use foreign models, compared with 65% of everyone else. Russian models show the opposite pattern: 14% vs. 28%.
Within groups: 248 and 292 people
What people use overall
% of AI users who chose each option (multiple answers allowed)
The larger the company, the less likely it is to choose foreign models: 86% of respondents at small companies use them, compared with 71% at large ones. Russian models show the opposite pattern: large companies use them more often than small ones (27% vs. 19%). Another 10% of respondents don't know which models their company uses.
05What brings AI out of the shadows
The more systematically a company rolls out AI, the less shadow AI there is
Where AI hasn't been rolled out, 55% of users regularly work from personal accounts. Where it's been rolled out centrally, that number drops to 27%.
Not rolled out
Partly
Centrally
Half as much shadow AI
The difference shows up at every step: the more a company invests in its AI rollout, the less often employees turn to personal accounts.
“The most motivated employees found the new tool on their own, figured it out and pay for the subscription themselves so they can work faster. All companies can do now is catch up with their own people.”
AI moved in on its own. Companies haven't noticed yet
88% of employees already use AI, yet only one in four says their company has formalized it with tools and policies. More than half of AI users handle work tasks through personal accounts, most often on foreign models. And where a company provides shared tools and policies, employees turn to personal accounts half as often. The question is no longer whether to adopt AI, but how quickly you can give employees an official tool and clear rules for using it.
How we collected the data
Who we surveyed
From June 30 to July 18, 2026, we surveyed employees of Russian companies. The AI section includes responses from 798 people: 104 from small companies, 311 from mid-sized and 383 from large ones. Respondents weren't just developers: 341 came from engineering and IT, 108 from operations, sales and HR, 106 from marketing and support, 103 from training and 33 from finance. Questions about tasks, models and personal accounts were asked only of AI users, about 708 people.
How to read the numbers
Our sample includes more digital teams than the market average, so treat the percentages as an upper-bound estimate. Comparisons between groups show correlation, not causation. All percentages are based on the people who answered each specific question.
We excluded some responses that failed our quality checks. This shifts the final percentages by no more than 1.5 percentage points. Some responses in the large-company segment were collected within individual organizations.
Using the data
Feel free to use the numbers and charts with a link to the Pachca study. For comments, email team@pachca.com.
This is just part one
Part two covers how companies roll out AI agents: the approaches they take, how deeply agents are built into everyday work and how companies solve the problem of sharing context. We'll email it to you as soon as it's ready.
Connect AI agents to Pachca
This study was conducted by the team behind Pachca, a team chat app for modern teams. We're building a workspace where AI agents work alongside employees, inside the company's official infrastructure.