During the London leg of its Agentforce tour, Salesforce reiterated its vision for the agentic enterprise, to enable customers to integrate the technology into their workflows in every corporate function. This means integrating Agentforce into the entire business from sales and commerce through to marketing, and into every app, ensuring predictability and accuracy through solid guardrails.
Salesforce UK and Ireland CEO Zahra Bahrololoumi said the CRM giant has focused on customers’ needs to design and create Agentforce 360, a platform with a unified vision to connect end users, employees, and operations, utilising AI agents. Leveraging contextual information is essential to drive value in agentic AI, but this entails giving agents permission to access every corner of the corporate data estate. Because of this, guardrails are crucial to control and manage the all-important governance aspect, which has a lot of bearing on cybersecurity.
Go deeper with GlobalData
Access deeper industry intelligence
Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.
New tools such as Agentscript are helping with this: during high-stakes situations it is essential to ensure that the agent is going to display predictable behaviour every single time, which involves designing and giving very clear rules similar to we would do with a small child; creating firm guidelines and guardrails becomes paramount.
Agentscript provides a scripting language enabling users to establish guardrails with natural language to ensure consistent adherence to rules. This is especially important during delicate conversations such as giving a financial advice or supporting a customer who is unhappy with the service. Intelligent Context, a new feature within the Data 360 (formerly Data Cloud) platform, helps with this by providing AI agents with access to both structured and unstructured data.
Another tool, Agentforce voice, allows users to apply voice to conversations with customers, adding personalised context for the agents.
Customer presentations
The event was heavy in customer presentations with a line-up prominently featuring Pandora, a jewellery maker that operates in more than 100 countries. The company has built an agent called Gemma, powered by Agentforce technology, which is always available on the website. Gemma is now handling 60% of all routine customer support queries, leaving the most complex ones for the human team, with more time in their hands to resolve tough problems.
US Tariffs are shifting - will you react or anticipate?
Don’t let policy changes catch you off guard. Stay proactive with real-time data and expert analysis.
By GlobalDataGemma is helping Pandora go from providing reactive support to customers to offering proactive services, guiding them in their journey, giving personalised recommendations and real-time data, not just on products, but also on the customers themselves. The company aims to combine the speed of online shopping with the personal experience of real life. For example, Gemma is aware of the shopper’s buying history and is able to discern what they have bought in the past, to refine recommendations. The agent is connected via data Salesforce’s Customer 360 to all customer information and the Pandora product information. It can also help users with queries about nearby Pandora stores and even book repair appointments for them.
Another example included the UK’s first policing digital worker, Bobbi, also powered by Agentforce, which can render efficiency gains for human staff to handle more sensitive queries and provide greater public access to non-emergency information. It leaves operators free to take 999 emergency calls.
Challenges ahead
Salesforce has completely pivoted around its agentic AI drive, since it introduced Agentforce in 2024. However, most large enterprises are still grappling with a huge technical debt and this is an important barrier to deploying agentic AI tools. The most notable challenge of all involves data cleansing, integration and governance. This is significantly slowing down adoption of agentic AI technologies in production environments, including Agentforce.
Despite this, Salesforce is demonstrating solid momentum with Agentforce, as shown in its latest financial Q3 FY2026 results, delivering a revenue forecast which topped estimates and revealed Agentforce and Data 360 (formerly Data Cloud) as strong drivers of sales. The company has repeatedly acknowledged that the main challenge remains taking AI projects from proof of concept to production, and the proliferation of DIY projects that render poor results, hampering enterprise adoption in spite of the remarkable practical advances of agentic AI.
