Agentforce World Tour London Recap

Salesforce Wants You to Become an “Agentic Enterprise”

This was undeniably the day’s key message. Salesforce is betting big on the “agentic enterprise, a business where people and intelligent AI agents work together”.

The Agentic Enterprise represents a fundamental shift in the way the world does business – leveraging AI agents, not just to automate existing workflows, but to transform the entire enterprise.

Salesforce.

Most of the hour-long keynote hosted by President & CEO of Salesforce UK & Ireland, Zahra Bahrololoumi CBE, expanded on this vision. It’s free to watch in full on YouTube.

Although the keynote had a bit of buzzy language, it was refreshingly honest.

Firstly, in the concession that Salesforce itself is “well on our way to becoming an agentic enterprise,” but not quite there yet. And secondly, by acknowledging that using agentic AI properly is not an easy or fast transition.

Salesforce Says Agentic AI Is No Longer a Pilot or Prophecy

We saw at Agentforce World Tour London that agentic AI is moving out of experimentation and into real business workflows. There were several real-world case studies shared throughout the day:

• Thames Valley Police introduced “Bobbi”, the UK’s first police AI agent, which handles ~200 general public interactions a day (resolving 45% without human intervention) and has saved 3,200+ operational hours.
• Canada Goose has agents handling routine service interactions (orders, returns, product questions) end to end, freeing teams for complex cases.
• Formula 1 showed how they’re building and launching campaigns in minutes, with agents running targeting and optimisation, and humans owning the creative and strategy.
• Adecco runs multilingual agents, speeding up candidate and client interactions.

In one session that Maria, Tap CXM’s Salesforce Marketing Cloud Developer, attended on the day, it was posited that agentic AI has the potential to be a bigger shift than social media.

Social media changed how we communicate. Agentic AI changes how we work – at least that’s the plan.

So far, though? Well, Maria reported that Salesforce presented an interesting stat: 94% of organisations aren’t seeing ROI from AI.

That didn’t surprise our CEO, Andrew, who also attended Agentforce World Tour London.

While the Agentforce case studies on display were undeniably slick, most practical usages still seem to be in customer service rather than business intelligence. It’ll be interesting to see more use cases of AI agents enhancing coding or accelerating learning. That’s something we’re starting to explore internally.

Andrew Haines

Quick Explainer: What Is Agentforce?

Agentforce is the platform behind Salesforce’s “agentic enterprise” vision and the star of this Agentforce World Tour series.

It’s not one agent. It’s a platform for building autonomous agents that can:

• Understand context
• Reason through a task
• Take action, calling tools and datasets
• Work across systems to ‘get it done’
• Involve humans for important decisions

Importantly, Agentforce is only one player in a three-part agentic enterprise stack. It’s part of a new “Headless 360” layer which, together with existing Customer 360 and Data 360, forms an agentic-ready ecosystem.

The Big Announcement: Headless 360

Headless 360 exposes the Salesforce platform – data, workflows, business logic, permissions – as APIs, MCP tools, and CLI commands.

Agents get what they need without touching a browser.

This is Salesforce preparing for a world where people don’t always ‘use Salesforce’ by logging into Salesforce.

Instead, agents will use Salesforce on their behalf. They’ll be embedded within the tools where work already happens.

Headless 360 is what gives the tools access to Salesforce to make this possible.

“Headless 360 is a system that allows any agent, any service, and any builder to create [user experiences]. And it’s incredibly powerful. If you think about it, Salesforce, for 27 years, has built technology and products to uplift humans. Salesforce Headless 360 is how we do the same thing for agents.”

Joe Inzerillo, President, Enterprise & AI Technology, Salesforce.

Salesforce’s recent acquisition spree (Slack, Contentful, Fin, Informatica, Cimulate, and Qualified, to name a few) no doubt feeds into this long-term vision.

But it’s also geared to work with non-Salesforce agents and platforms. For example, we saw Claude and ChatGPT configured to access Salesforce data during the keynote.

Tap’s Takeaways From Agentforce World Tour London 2026

Salesforce is making a serious push for companies to move from AI as a feature to AI as an operating layer.

We’re excited about that direction. We’re already working with Agentforce capabilities and see clear potential for clients.

So far, it’s typically starting with agents that handle repetitive, high-volume work. Customer support is an obvious one. We’ve also built a few other prototype agents for internal challenges.

It’s a welcome circuit-breaker, even if it’s just a start. Almost every organisation today – enterprise or otherwise – is under pressure to improve performance but constrained by resources.

But it is just a start. There are a few things you need to keep in mind.

Today’s Agents Are Not the Final Evolution

As agent capabilities and user confidence evolve (likely in step), the use cases will get more sophisticated.

Walk Before You Run

The marketers that get the most benefits from agentic AI won’t necessarily be the fastest movers. They’ll be the ones who’ve sorted their data and kept people in the loop where it counts.

Human interaction is still needed to protect organisations when it comes to governance, guardrails, and decision-making.

Don’t Expect an Instant Fix

You need to do some groundwork. Agents aren’t plug-and-play. They need a solid data foundation, the right tool connections, governance, good prompting, and humans on critical decisions.

Get those right, and they’re powerful. Skip them, and you automate a mess.

Consider the Long-Term Workforce Impact

You’ll see a lot of cost-saving KPIs in the AI space, especially ‘person-hours saved’. No one is being bold enough to outright say this has led to headcount cuts. The optics are understandably bad – and the long-term outcomes tend to be worse.

AI doesn’t replace people. It’s a tool they use to do better work.

Don’t Become a Hammer In Search of Nails

Salesforce themselves highlighted that people are still confused about what AI should be doing for them. C-Suites are asking what their organisations are doing to adopt ‘it’ without having a clear direction or well-defined challenges.

Look For Use Cases Beyond Cost

We’ll slowly move away from developing agents to save costs towards delivering a better customer experience. By 2027, we expect half of all new agents being developed will not focus solely on saving costs.

Operating Models Must Change

Whether it’s hiring from the new generation of AI-native graduates or rethinking how customer-facing decisions get made, agentic AI will gradually make its mark on the way work gets done.

If you’re interested in exploring Agentforce capabilities in more depth, we’d love to chat. Feel free to get in touch any time.