Governance Lags Agentic AI Adoption in the UK
An orchestration and governance gap is emerging as enterprises deploy AI agents, according to recent research from Salesforce. The ‘2026 Connectivity Benchmark Report’ reveals that organizations have an average of 12 agents, a figure projected to increase by 67% over the next two years.
Key Findings
This report, the 11th iteration produced in collaboration with Vanson Bourne and input from Deloitte Digital, surveyed 1,050 IT professionals across multiple countries, including the UK, US, Ireland, France, Germany, the Netherlands, Australia, Singapore, Hong Kong, and Japan. The UK cohort comprised 100 respondents.
Notably, 89% of organizations in the UK and Ireland already deploy AI agents; however, half of these agents remain compartmentalized into silos and are not utilized effectively at an enterprise level. Alarmingly, only 54% of organizations maintain a centralized governance framework for their agents.
Challenges and Concerns
Respondents expressed concerns regarding the potential complexity introduced by agents, with 75% worried that agents might complicate operations without providing sufficient business value. On average, enterprises are managing 796 applications, of which only 33% are integrated with one another.
Furthermore, a significant 97% of the organizations identified barriers to utilizing data for AI use cases, with 35% citing outdated IT architecture and infrastructure as top blockers due to data silos and disconnected systems.
The Future of AI Adoption
Beena Ammanath, Global Deloitte AI Institute leader, noted that AI adoption speed has outpaced predictions, with 84% of enterprise CIOs believing that AI will become as crucial as the internet for their businesses. Currently, 40% of organizations report having deployed autonomous agents to augment existing processes, with another 41% planning to implement them within the year.
Transitioning to Multi-Agent Systems
During a recent press briefing, Kurt Anderson, Managing Director and API Transformation Leader at Deloitte Consulting, cautioned against the risks of shadow AI. He emphasized the importance of empowering individuals who can effectively deploy knowledge while maintaining security and reliability.
Andrew Comstock, Senior Vice-President and General Manager of MuleSoft, stressed that the success of an agentic enterprise lies not in the number of agents deployed but in their overall effectiveness. He highlighted the need for a unified foundational framework to ensure that multi-agent systems are safe, reliable, and scalable.
AI in Sales
Additional research from Salesforce indicates that agentic AI software is becoming a vital tool for sales professionals in the UK, ranking among the top three sales techniques for 2026. Those who effectively utilize AI agents are 1.7 times more likely to meet their sales targets.
While the average UK seller spends 41% of their time selling, younger sales staff, particularly Gen Z, report spending only 35% of their time on sales activities. Furthermore, 90% of surveyed sales professionals are utilizing some form of AI, with 46% having used agents.
Conclusion
As organizations strive to embrace the agentic AI revolution, the need for robust governance and integration becomes increasingly clear. Companies must navigate the complexities of adopting AI while ensuring that their processes support sustainable growth and value creation.
In summary, while the potential for AI agents is immense, organizations must address existing challenges to unlock their full capabilities in the enterprise landscape.
