Agentic AI gets real at Salesforce Dreamforce

The energy and scale of this year’s Salesforce Dreamforce made it feel like you were in a city, not at a technology conference.

With tens of thousands of people, over 1500 sessions, dozens of stages and a constant buzz about the next big leap in AI, this was the place to be for customer-centric innovation.

And the word that defined it all? Agentic.

Real tools, real results: Salesforce Agentforce in the spotlight

For conference-goers at San Francisco’s Moscone Center, Salesforce Agentforce was a recurring theme.

This latest evolution of the Salesforce AI ecosystem promises AI agents that don’t just predict or automate but also act.

These are not theoretical models or future-looking promises. They are real tools, producing real results — and practical use cases built within the Salesforce platform were very much in evidence this year. Salesforce showcased functioning agentic systems embedded across its ecosystem. Sales Cloud, Service Cloud, Revenue Cloud, Marketing Cloud and Tableau are all increasingly underpinned by Agentforce.

It’s an ambitious vision: A single, interconnected platform where AI agents can pull data, take actions and support decisions autonomously.

Data remains the unseen hero

Part of this platform is Agentforce Vibes, which provides a quick way to create AI agents. “Unleash developer productivity — simply describe what you need, and your AI-powered coding partner writes, tests and maintains the code,” says Salesforce.

That’s an impressive achievement, but you can’t mention AI or Agentforce without talking about data. As we saw throughout the event, none of the magic happens without strong data foundations.

Everyone wants that perfect, AI-enabled 360 customer view, but if your data is messy or incomplete, you’re just teaching your AI to make poor decisions faster.

That’s where we, as consultants and technologists, come in. Before you can tap into agentic AI, you need to get your house in order by consolidating systems, cleaning data and ensuring scalability.

In short, don’t think of Agentforce as step one. It should be the final piece of a well-built foundation.

Avoiding the “AI everywhere” trap

If you’ve been to any technology event this year, you’ve probably heard “AI” at least a hundred times before lunch. But at Dreamforce, we noticed a subtle shift in the conversations. The questions are different, focusing more on implementation than experimentation. The excitement is still there, but it’s tempered by pragmatism.

Customers are starting to become wary. Every vendor promises AI, but how do you know what’s right for your organization? Do you pick one platform or a combination of specialized tools that talk to each other?

One idea that caught our attention was the emergence of agentic integration layers — AI agents that talk to other agents across platforms, replacing traditional application programming interfaces. This means systems won’t have to be manually integrated, as they can understand each other through intelligent negotiation.

It’s an exciting concept but also a reminder that you need to move strategically, not reactively. Jumping on every AI trend without a roadmap leads straight into what we call the agentic pilot trap — where projects sound promising but never scale.

• ALSO READ → Escape the agentic AI pilot trap with NTT DATA and Agentforce

New ways of thinking about ROI on AI

While some executives are bullish on AI investments, many struggle to define what success looks like, and it’s easy to see why. Traditional ROI models based on cost savings, revenue growth can’t always capture the full value of AI — like saving a customer two hours of waiting time.

In countries dealing with high unemployment rates, there’s also a social context — AI isn’t about replacing contact center agents but rather about augmenting them.

It’s about productive coexistence. Let your AI agents handle routine queries so human agents can focus on the complex, emotional conversations that really build customer loyalty. That’s how you increase customer lifetime value and wallet share.

Looking ahead to 2026

So, what’s next? For Salesforce clients in the Middle East and Africa, 2026 will be a year of choices — not just about which technologies to adopt but also about how to connect them meaningfully.

The organizations that make clear progress won’t necessarily be the ones with the most advanced AI but those that align AI with real-world outcomes such as faster service, smarter marketing and happier customers.

AI won’t transform your organization by itself. But when you get your data right, define your use cases clearly and let your agents talk to each other, the magic happens.

This article was co-authored by Jenine Binos, Senior Sales Manager: CX & Design at NTT DATA in the Middle East and Africa.