Governance lags agentic AI adoption within the UK, says Salesforce

An orchestration and governance hole is rising as enterprises deploy AI brokers, in accordance with analysis from Salesforce. In its Connectivity benchmark report, Salesforce-owned Mulesoft discovered that the organisations surveyed have 12 brokers on common, with the quantity projected to climb by 67% over the subsequent two years.

That is the seller’s eleventh iteration of the report, produced in collaboration with Vanson Bourne and with enter from Salesforce’s Mulesoft associate Deloitte Digital. It surveyed 1,050 IT professionals from the US, UK and Eire, France, Germany, the Netherlands, Australia, Singapore, Hong Kong and Japan between October and November 2025. The UK cohort comprised 100 respondents.

It discovered that 89% of the UK and Eire organisations already deploy AI brokers, however that half of the brokers are compartmentalised into “silos”, so should not meaningfully deployed at an enterprise stage. Solely 54% of organisations have a centralised governance framework with formal oversight for his or her brokers. Some 94% of the respondents agreed that AI brokers would require new IT architectures to grow to be extra API-driven to succeed.

Enterprises are utilizing 796 functions on common, with 33% of them built-in with one another, whereas 75% of the respondents are involved that brokers will introduce extra complexity than enterprise worth.

The bulk (97%) of the organisations stated they’ve limitations to utilizing information for AI use instances, with 35% figuring out outdated IT structure and infrastructure resulting from information silos and disconnected programs as being the highest blocker. 

Within the foreword to the report, Beena Ammanath, International Deloitte AI Institute chief, stated: “AI adoption velocity has surpassed predictions from even a yr in the past, with 84% of enterprise CIOs believing that AI will probably be as necessary to their companies because the web.

“Now, on the cusp of the subsequent period of AI – the agentic enterprise – 40% of organisations report having deployed autonomous brokers to enhance present processes and groups, with one other 41% planning to deploy brokers throughout the subsequent yr.”

In a press briefing on transferring from single agent deployments to multi-agent programs, at which the analysis behind the Connectivity benchmark report was offered, Kurt Anderson, managing director and API transformation chief with Deloitte Consulting, expressed warning in regards to the hazard of shadow AI.

“There is a vital management precept of placing information within the fingers of the individuals who can deploy it, who’ve one of the best info to decide and construct an agent,” he stated. “What’s totally different about this [agentic AI] revolution is we’re embracing that information and tech is decreasing the bar to constructing your personal capabilities…We are able to put guard rails and transparency and safety in place and so get information into the brokers.

“That half is basically cool in contrast with a previous method that required a extra top-down, solid-hand kind of governance. However we do must be taught from the previous, and ensure we’ve got the proper governance. We all know what occurs whenever you let particular person contributors unfastened when it comes to constructing programs that lack safety or reliability or run up an enormous licensing invoice.”

In a publish pointing to the report, Andrew Comstock, senior vice-president and basic supervisor of MuleSoft, stated: “The true success of an agentic enterprise isn’t discovered within the sheer variety of brokers deployed, however the general effectiveness of these brokers. We want to consider how they’re found, ruled and orchestrated to work collectively.

“As we transfer into this multi-agent period, the function of IT is evolving from managing silos to constructing a unified basis because the central management airplane that enables multi-agent programs to be protected, dependable and scalable.”