How piloting Agentforce with a customer-zero mindset is transforming sales and delivery at Accenture.

Rajendra Prasad: Operating as customer zero is our number one credential and number one priority. It proves we know the technology. When I talk to CIOs, CTOs, CMOs, CHROs, finance leads — no matter the title, every functional requirement connects back to technology. And every conversation includes AI. I'll be more authentic when I can explain how I set up Agentforce.

As an example of this value in action, at Accenture, we have our own structure, our own ontology. We define Marketing Units (MUs) as a geography. But out of the box, the MU field was not recognized by Agentforce, so we had to customize it. When I can detail what changes we made to the database layer or product experience layer to work within our processes, I can speak directly to how their enterprise can optimize too, and also how Agentforce becomes increasingly effective as you work with it.

Rajendra Prasad: I strongly believe that Salesforce is unique because of the fundamental architecture of the platform. I’m going to use a technical term: It’s the vertical integration of the technology stack. What does it mean? That you can start fast. This is because when an enterprise's sales DNA is tightly linked to Salesforce, the architecture flows from the compute layer, to model layer, to interaction layer, to the agent front-end in a vertical, integrated process.

Here is how that is already coming to life at Accenture: In our sales process, we classify and tag each sale into Service Groups so we can report revenue by service line. Those cell tags are coded into the database. But what about when someone creates a new opportunity and wants a tag that doesn’t exist? Historically that meant functional design, reviews, application changes, testing, and implementation.

With Agentforce, the AI agent can either guide them like a trusted coworker or automatically take on the tagging workflow — from creating the sales tag to capturing the data. Something that was complex is now simple. That simplicity means more time and more revenue.

Sara Porter: Technology adoption is a people story, and it goes back to having the right people in your pilot. At Accenture we have the benefit of scale. Our pilot is 2,200 users, which at some companies, would be huge. We have 55,000 sellers, so 2,200 is a fraction. We were intentional about creating the right cross-section with different geographies, languages, roles — leaders, managers, individuals. Our pilot group is well respected across the org, so if they’re using Agentforce and advocating for it, that drives adoption. From our CEO Julie Sweet down, everyone is moving towards operating in an Agentic Enterprise, particularly in our sales process.

You’re also right about it sounding extended – we’ve been in pilot for six months, which may seem long. Yet, when I think about our initial deployment of Agentforce Sales in 2013, that was a six-month pilot, too, but it was much more of a waterfall approach. When piloting Sales Cloud, I told the team, “Go use it, give us feedback.” But Agentforce is different, because it learns in real time. It’s agile – you just give the agent’s response a “thumbs up” or “thumbs down” and it self-corrects immediately.

Rajendra Prasad: We were among the first to go live with Agentforce and are actively infusing it into workflows. An example of our employees augmented with agents is deal coaching.

We’ve always had managers coach high-value or intricate deals, but now we also have an Agentforce sales coach that stays with a seller throughout the lifecycle — from request for proposal (RFP) to real-time client interactions — and guides next steps. Agentforce, together with our best-in-industry sellers, is changing how we win.

Another example: Behind the scenes, I have a data supply chain that flows from the data’s origin, through a business intelligence (BI) layer. From the moment a sale is created, to each keystroke, to batch processing, to run processing, to front end, to the name of the employee who sold the deal, we’re creating data — and it’s mission-critical that all that data finds its way into Salesforce. 100% of our deal proposals, 100% of opportunity management, everything generates data and every piece of data flows through our Sales Cloud org.

Now by applying Agentforce on top, I can query, “How do I make more business? What areas are selling better? Where are we losing deals? How do we make the impact so these are all the critical decisions, well past the selling phase and into delivery?”

Sara Porter: The biggest exit criteria for scaling org-wide is accuracy, but really we’re driving toward three goals. First, maximizing the quality and accuracy of Agentforce. Agentforce isn’t a preset building block, it’s more like a new employee that needs onboarding. That means engaging with it in multiple ways during the pilot and tracking the questions we ask.

Second, selecting prompts and use cases that are actually helpful across different groups. We already have plenty of tools; Agentforce needs to be about removing friction.

And third, proving business value. For us this is predicated on time savings. We ask the pilot group daily, “Is Agentforce reducing your time to get things done? Is it improving your output? How could it do more?” Even if a seller only uses Agentforce to update one CRM field, we measure the time saved and then ask, “Can we apply this automation across dozens of fields in Opportunity Management?” That’s where the real power comes in.

Full deployment will look like a hockey stick. We’ll grow from 2,200 to 16,000 users next, then to all 55,000. And because Agentforce is already saving time, we’re confident that each phase will add value.

Rajendra Prasad: This might be surprising, but architecting your underlying data fabric is more important than configuring agent-to-agent interactions. Salesforce, and the strength of the built-in structure, gives me cohesion in one system. Not all solutions approach this A2A relationship the same.

Simply put, an “A to B to C to D” workflow among four agents is very easy. The hard part is the data flow between them. Without the right architecture, your output from A to B might be fine, but from A to D – that relationship will break. Salesforce is designed in a way that Agentforce interactions can be brought to life much easier.

Both the underlying data and the data stratification capabilities were fundamental to Salesforce even before Agentforce. When Agentforce came, I was very confident that the data fabric was ready to use, which would directly correlate to the speed at which I could go. In fact, this is what we should be talking to all of our clients about.

When I talk to a CIO about implementing Agentforce, I ask them to show me their BI layer on Salesforce. If I understand that the BI layer is 60–70% automated, then the answer is very simple. Agentforce can just plug in – fast.

Sara Porter: We have been on a reinvention journey at Accenture and each of our functions are driving unbelievable transformations in marketing, legal, finance. As an Agentic Enterprise, it’s no longer, “Marketing, you're the flute section. Here comes legal on the violin, and finance, with the clarinets.” Now, we’re one AI symphony because Agentforce is actually bringing that collaboration together in a way that we’ve never been able to do before. It's using agentic AI to make every process stronger. We drew a picture of humans and digital agents as part of the same team. That's where AI should be going, and that's what I'm excited about.