Companies Are Adopting AI Agents Faster Than They Can Govern Them
Just 24% of companies have a company-wide AI usage policy and only 25% measure AI’s impact with clear KPIs, even as more than half are running or actively piloting AI agents.
AI agents have moved quickly from experimentation into active business use. Companies are now focused on how to govern agents, preparing employees to work with them and determine whether they’re creating meaningful value.
But moving fast on adoption isn't the same as capturing value from it. A company can have agents live in the business and still have no formal policy governing how they're used, no training plan for the employees working alongside them, and no clear way to measure whether any of it is paying off.
About the Liferay 2026 Agentic AI Adoption and Governance Report
The Liferay 2026 Agentic AI Adoption and Governance Report looks at where companies actually stand on agentic AI: how far adoption has gone, what's holding back the value companies expected and which industries are furthest ahead and behind.
Liferay partnered with the third-party survey platform Pollfish to survey 500 full-time employed U.S. adults who have a direct hand in their company's AI decisions, evaluation, or day-to-day use.
Key Findings of the Liferay 2026 Agentic AI Adoption and Governance Report
- 54% of companies are running AI agents in production or actively piloting them today.
- AI agent adoption varies widely across industries, from technology at 72% to education at 24%.
- 42% of companies say better accuracy is what AI agents need most to become more useful.
- Data privacy is the top concern about AI agents overall, cited by 21% of companies.
- Security or privacy concerns (30%), cost (29%), and lack of training (27%) are the leading barriers to getting more value from AI.
- Only 24% of companies have a company-wide AI usage policy.
- Only 25% of companies measure AI's impact with clear KPIs.
Most Companies Are Already Running or Piloting AI Agents
More than half of companies, 54%, are already running or actively testing AI agents. That includes 28% with agents in production and another 26% in active pilots or testing. 16% are considering deploying AI agents within the next 12 months. 13% are interested with no active plans. 9% have no plans. 8% are not sure.
Only 6% of companies say they're not comfortable with agent autonomy yet. 21% are comfortable letting agents take actions automatically within strict rules, the largest single group. 17% are comfortable with full autonomy for selected workflows.
Enterprise AI Adoption Varies Widely by Industry
Technology's adoption rate (72%) is roughly three times that of education's (24%), the widest gap between any two industries in the survey. Five industries: technology, healthcare, transportation, manufacturing and financial services, run or pilot AI agents at a higher rate than the 54% survey average. Professional services matches that average exactly, while retail, real estate, government and education all fall below it.
29% of government organizations are interested in AI agents but have no active plans to implement them, the highest share of any industry.
Data Analysis and Customer Support Top the List of AI Agent Use Cases
Top use cases companies picked for where AI agents would be most useful: data analysis/reporting (35%), customer support (32%), IT service desk (21%), finance operations (19%), HR operations (18%), website personalization (18%), content production workflows (18%), internal knowledge management (18%), sales development/lead qualification (16%), software development (15%), marketing campaign execution (14%), legal/compliance (9%).
Companies already using AI report similar priorities today: data analysis (44%) and writing or editing content (44%) are the two most common current AI use cases, followed by research (42%).
Accuracy and Security Are What AI Agents Need Most
Top requirements companies name for AI agents to become more useful: better accuracy (42%), stronger security controls (36%), better employee training (35%), clear human approval workflows (32%), access to trusted internal data (25%), integration with existing business systems (24%), clearer governance policies (23%), clearer source citations (20%), clearer ROI measurement (18%), executive sponsorship (17%).
Data Privacy and Employee Trust Lead AI Agent Concerns
Data privacy ranks as the top concern about AI agents at 21%, followed by employee trust at 17% and security risk at 13%. Incorrect actions and cost rank lower, at 11% and 10%, respectively.
The emphasis on employee trust might also help explain why AI training is the most common planned AI investment for the next 12 months at 39%, ahead of AI agents specifically at 21%.
Government Organizations Lag in AI Agent Adoption as Privacy Concerns Remain High
Only 38% of government organizations are running or piloting AI agents, while 43% cite data privacy as their top concern and just 14% report having an organization-wide AI usage policy.
14% of government organizations are also not comfortable with agent autonomy yet, the second-highest rate of any industry after education (15%).
Security, Cost, and Training Outrank Unclear ROI as Barriers to AI Value
Top barriers to getting more value from AI overall: security or privacy concerns (30%), cost (29%), lack of training (27%), legal or compliance concerns (22%), employee resistance (21%), poor data quality (21%). Unclear ROI ranks last among the options tested, at 12%.
Financial Services Leads on AI Governance, but Compliance Concerns Persist
More than half of financial services companies, 56%, are already running or piloting AI agents. The industry also leads every other sector in formal AI governance, with 38% reporting a company-wide AI usage policy, compared to 24% overall.
Even with that stronger policy foundation, 38% cite legal and compliance concerns as their top barrier to getting more value from AI, the highest share of any industry.
Most Companies Lack Clear KPIs to Measure AI's Impact and ROI
25% of companies measure AI's impact with clear KPIs. 23% are planning to. 22% measure it somewhat, but inconsistently. 16% are not sure. 13% do not measure AI's impact at all.
How Enterprises Can Close the AI Governance Gap
The data shows where AI value stalls: incomplete governance policies, gaps in employee readiness and inconsistent measurement.
The next phase of enterprise AI will require a clear operating model that connects tools, data, permissions and human oversight. Organizations need to determine which data agents can access, what actions they can take, where human approval is required, and how success will be measured. Without that foundation, adding more agents may simply introduce more complexity.
Liferay helps enterprises establish that foundation through Liferay AI Hub, which lets teams deploy AI agents with fine-grained permissioning, so agents can only access the information a given user is already authorized to see. Learn more about Liferay DXP.
AI agents help companies move faster when employees know how to use them and leaders set clear expectations for where they belong. Companies need to decide what data agents can access, which actions require human review and how they'll measure whether the technology is helping. Those decisions can't wait until agents are already spread across the business. Building that foundation early gives teams more confidence to use AI in meaningful ways without creating unnecessary risk or complexity.
Liferay Co-Founder & CMO
Frequently-Asked Questions About the Liferay 2026 Agentic AI Adoption and Governance Report
What are AI agents? AI agents are AI systems that can complete multistep tasks, use tools or internal systems and take actions with limited human input. Only 6% of companies say they're not comfortable with agent autonomy at all.
How many companies are using AI agents? More than half. 54% of companies are running AI agents in production or actively piloting them today, according to the Liferay 2026 Agentic AI Adoption and Governance Report. Another 16% are considering it within 12 months, and only 9% have no plans to use them.
Which industries have adopted AI agents the fastest? Technology (72%) and healthcare (66%) have the highest AI agent adoption rates, followed by transportation and logistics (64%). Government (38%) and education (24%) have the lowest.
Where do companies think AI agents would be most useful? Data analysis and reporting (35%) and customer support (32%) top the list, followed by IT service desk support (21%) and finance operations (19%).
What do AI agents need to become more useful? Better accuracy (42%) and stronger security controls (36%) are the top requirements companies name, followed by better employee training (35%) and clear human approval workflows (32%).
What's the biggest concern companies have about AI agents? Data privacy. It's cited by 21% of companies overall, ahead of employee trust (17%), security risk (13%) and agents taking incorrect actions (11%).
How does government AI agent adoption compare to other industries? It's lower. Only 38% of government organizations are running or piloting AI agents, compared to 54% overall. Government also reports the highest concern about data privacy (43%) and one of the lowest rates of organization-wide AI usage policy (14%, versus 24% overall).
What's the biggest barrier to getting more value from AI? Security or privacy concerns (30%) and cost (29%), followed by lack of training (27%) and legal or compliance concerns (22%). Unclear ROI ranks last among the barriers tested, at 12%.
Is financial services ready for AI agents? Mostly, but not without friction. Financial services leads every industry on formal AI governance, with 38% of companies reporting a company-wide AI usage policy. But 38% also cite legal and compliance concerns as their top barrier to AI value, the highest of any industry.
Are companies measuring AI's return on investment? Not most of them. Only 25% of companies measure AI's impact with clear KPIs. 23% are planning to, 22% measure it inconsistently, and 16% aren't sure whether their company measures it at all.
What are companies investing in for AI in the next 12 months? AI training. It's the most common planned investment at 39%, well ahead of AI content tools (26%), AI workflow automation (25%), and AI agents specifically (21%).
Survey Methodology
The Liferay 2026 Agentic AI Adoption and Governance Survey was conducted via Pollfish among 500 full-time employed U.S. adults with a direct hand in their company's AI decisions, evaluation, or regular use. The survey assessed AI agent adoption, comfort with agent autonomy, barriers to AI value and planned AI investment over the next 12 months.
