The most common barriers amongst organizations that use AI agents include:
lack trust in AI outputs
cite employee resistance
struggle with integration challenges
cite cost and compliance concerns
Underlying concerns across all organizations reinforce those barriers:
worry about job displacement
cite security and compliance risks
remain uncertain about return on investment
The barriers organizations face are not uniform. Each sector encounters a different primary blocker as AI moves from experimentation into operational scale.
Legal
Legal organizations report the highest concerns around both trust in AI outputs and data integration (60%), underscoring how high the accuracy threshold becomes when professional liability is involved.
Banking
Banking organizations cite data integration as their single largest barrier (45%), reflecting the weight of legacy systems and siloed infrastructure.
Healthcare
Healthcare organizations report comparatively lower deployment pressure, but compliance and regulatory risk remain major constraints (33%).
Retail
Retail organizations report the fewest structural barriers overall, helping explain why conviction and investment intent are accelerating quickly within the sector.
One of the clearest indicators of maturity is how organizations handle failure.
Technology and Banking organizations are the most likely to have paused or canceled AI initiatives, at 49% and 42%, respectively, across all respondents. At the same time, they also report the highest awareness of Agentic AI, at 92% and 91%.
That combination is important.
The sectors experimenting most aggressively are also the most familiar with emerging AI systems and the most willing to abandon projects that fail to scale operationally.
Technology organizations:
Paused or canceled AI initiatives
Aware of Agentic AI
Banking organizations:
Paused or canceled AI initiatives
Aware of Agentic AI
BY CONTRAST
Legal and Healthcare organizations report both lower deployment and lower cancellation rates, suggesting fewer organizations are running enough initiatives to generate meaningful operational learning.
In practice, a willingness to pause or shut down projects is often a sign of experimentation maturity rather than poor execution.
Organizations running enough initiatives to generate meaningful operational learning are naturally more likely to identify projects that are not scalable, sustainable, or producing measurable value.
Lower cancellation rates, in context, can indicate under-experimentation rather than disciplined execution. As organizations scale AI across disconnected systems and workflows, friction compounds:
- Integration gaps reduce visibility
- Inconsistent outputs reduce trust
- Unclear ROI slows investment
- Governance requirements increase operational drag
That pressure becomes more visible in Midmarket organizations deploying AI agents, where compliance and regulatory concerns reach 39%, compared to 31% among SMBs and 33% among Enterprises, as scaling efforts begin to outgrow simpler governance structures.
Midmarkets
SMBs
Enterprises
Over time, coordination becomes the primary challenge. The organizations that scale successfully are building systems that can support AI consistently across the business.
RingCentral, the RingCentral logo, and all trademarks identified by the ® or ™ symbol are registered trademarks of RingCentral, Inc. Other third-party marks and logos displayed in this document are the trademarks of their respective owners.
© 2026 RingCentral, Inc. All rights reserved.