Escape the agentic AI pilot trap with NTT DATA and Agentforce

GenAI and agentic AI are now in every boardroom, every strategy deck and every pilot project. But the inconvenient truth is, many of these projects go nowhere.

MIT’s State of AI in Business 2025 report shows that 95% of organizations don’t see real returns from GenAI investments, and only 5% of task-specific tools ever make it to production.

The problem is that organizations keep falling into the “pilot trap” — launching proofs of concept that prove AI can do something, but not whether it can scale across the organization.

In the AI context, orchestration refers to the process of coordinating and managing multiple AI agents to achieve a complex task or workflow by integrating their capabilities, handling interactions and ultimately optimizing overall business performance.

Building a feature may deliver an impressive demo, but building an orchestrated system transforms the way you work. 

Why do GenAI and agentic AI pilot projects stall?

Pilot projects are useful because they prove feasibility. But they almost always stop at one-off use cases: a chatbot in customer service, a scheduling assistant in HR or a lead generator in sales.

The problem is that single agents don’t share context. They can’t coordinate, and they drop the ball when workflows require collaboration. What use is a chatbot that can answer FAQs but not escalate smoothly to a human sales agent? Or a lead generator that updates your customer relationship management (CRM) system but doesn’t sync with marketing campaigns?

This lack of coordination and collaboration leaves you stuck in demo territory. To get out, you need an adoption model built for orchestration — one that designs for coordination from the start.

The shift from apps to the agentic economy

Think about how we interact with technology. For the past two decades, apps have dominated our digital lives. You downloaded them, tapped through menus and filled out forms.

But this is changing as we move into the agentic economy, where intelligent AI agents replace apps as the main way people and organizations engage with technology. Instead of logging into an app, you interact with an agent that proactively handles tasks, makes decisions and coordinates on your behalf.

A powerful platform for this transition is Salesforce Agentforce, which connects AI agents across functions and industries to turn standalone AI tools into an enterprise-wide system of intelligence.

Importantly, the value of Agentforce is not limited to Salesforce only. It can also be extended to orchestrate AI agents on other platforms, to accelerate business value and impact by creating front-office to middle-office and back-office solutions.

Agentforce: The new nervous system of your organization

If your organization is the body, Agentforce is the nervous system. NTT DATA’s Smart AI Agent™ Ecosystem is how we strengthen it: not with one-off bots, but with an enterprise ecosystem — managed services, accelerators and best practices, a governed marketplace of agents and models, and secure, scalable infrastructure — that is built to deliver assured outcomes and responsible AI from day one.

With Salesforce Agentforce, our agents work as a coordinated team, enabling multiagent collaboration not just within your organization but also across supply chains, partners and even entire industries. This is where the competitive edge emerges: when agents collaborate as fluidly as humans do, but with machine speed and precision.

Think of a logistics firm rerouting around a storm. On Agentforce, a weather agent flags risk, a shipment agent replans routes, a customer communications agent updates cases and notifies affected customers, and a contracts agent initiates renegotiations. Human operators approve exceptions. That’s orchestration in action.

How to cross the divide from pilot project to scale

To help you activate this valuable orchestration layer, we’ve developed the EPAS adoption model:

• Evangelize: Get your leadership aligned on outcomes, not features. The question isn’t “Can we deploy a chatbot?” but “Can we reduce resolution times by 40%?”
• Pilot: Demonstrate multiagent coordination in one real workflow, with handoffs and human-in-the-loop, not just a single tool in isolation.
• Adopt: Make your data agent-ready. Move beyond simple unification into contextualized, governed orchestration of data across platforms.
• Scale: Establish repeatable orchestration patterns and interoperability, both within your enterprise and across your partner ecosystem.

The EPAS model derisks the jump from pilot project to production, helping you embed AI agents into processes where they move the needle.

Accelerating outcomes with Agentforce and Salesforce Data Cloud

When we pair Agentforce with Salesforce Data Cloud, the outcomes accelerate.

Telecommunications provider Vodafone España can now unify fragmented customer data, personalize campaigns, and launch them 80% faster. Brewing company Damm has reduced the time they spent on updating and improving their database of hospitality providers by around 70% — equivalent to 7,000 hours saved each year.

Systems integration, data unification and agentic orchestration fuel these competitive shifts.

Orchestrating agentic AI at scale

Many partners can implement Salesforce features, but few can orchestrate agentic AI at enterprise scale. This is where NTT DATA makes a real difference.

We’ve partnered with Salesforce for 25 years, assembling more than 4,700 certified experts and over 15,000 Salesforce and MuleSoft certifications to deliver at scale. We pair this talent with the full stack — cloud and data platforms, networks, IoT, security and governance — so your Agentforce rollout is safe, performant and auditable from day one. In client projects, we’ve reduced IT effort by up to 50% and costs by up to 45% while delivering orchestrated systems that actually transform how your business works.

We know you can’t just plug in a fleet of agents and hope for the best. Trust is key. If your customers and employees don’t trust these AI agents, adoption will be slow. That’s why we build responsible AI practices directly into Agentforce orchestration — to enforce transparency, auditability and fairness. This includes robust guardrails, clear human oversight for high-risk decisions and explainable AI actions.

Start with one workflow

The winners of the agentic economy will be those organizations whose AI agents are collaborating securely, responsibly and at scale.

The first step? Pick one workflow. Let NTT DATA and Agentforce prove the orchestrated outcomes there. Then, we’ll help you scale what works. It’s how you turn agentic AI into a reliable tool for measurable transformation.