Now, Near, and Next: How to build your AI Cloud ERP roadmap

In a recent episode of The Future of ERP podcast, we sat down with Ramesh Chougule, Vice President, SAP Practice Sales and Go to Market North America at Infosys, to talk about a question that a lot of organizations today are asking: Where exactly do we start with AI and cloud ERP, and how do we make sure we're building something that lasts?

Ramesh has spent 25 years working with SAP clients across hundreds of engagements, and he brought a clarity of thinking to this conversation that I think a lot of organizations need right now.

The hype around AI is loud. The question of what to actually do first is harder to answer. This blog pulls out the key themes from our conversation—including Ramesh's practical three-horizon roadmap—and his direct perspective on why the operational core of cloud ERP is more important in the age of AI, not less.

Is AI going to replace ERP?

We opened with the question that Ramesh told us has been driving client conversations for the past two years. His answer was unambiguous:

That dependency is the thread that runs through everything Ramesh shared. AI agents need governed, structured, reliable data to act on. Cloud ERP is where that data lives. Organizations that treat them as competing priorities will struggle; organizations that treat cloud ERP as the foundation for AI will compound their advantage.

What is the operational core—and why does it matter for AI?

Ramesh introduced a framework that Infosys has developed based on its extensive work with clients: a three-layer architecture for tomorrow's enterprise. Understanding it helps explain why foundation-setting matters so much right now.

On the relationship between the operational and strategic layers, Ramesh was direct:

In other words, the quality of an organization's AI-powered strategy is only as good as the data discipline in its operational core. Clean ERP data isn't a precondition for AI—it's a competitive advantage.

What does AI in cloud ERP look like in practice?

Ramesh gave concrete examples of where AI is already changing day-to-day operations—not as a future possibility, but as something his teams are seeing with clients now:

• A procurement manager receiving the best available contract price at the moment they're creating a purchase requisition—not after a manual lookup, but in real time, embedded in the transaction.
• A collections agent triggering an escalation automatically, without human intervention, to ensure receivables are received on time.
• Autonomous warehousing, where multi-agent AI orchestrates picking, packing, and shipping on the warehouse floor—avoiding stock-out situations and maximizing order fulfilment in real time.

On that last example, Ramesh pointed to Infosys's demonstration at SAP Sapphire 2026:

What connects these examples is the principal Ramesh kept returning to: AI delivers value when it's embedded in the process at the moment of execution—not sitting on top of the data, but acting within it.

What is the short- and long-term roadmap for AI and cloud ERP?

This is the practical framework I'd encourage any business or IT leader to take away from this conversation. Ramesh described how Infosys structures AI adoption roadmaps for clients across three time horizons:

On why the 'now' phase tends to be underestimated:

It's a point that's easy to gloss over when the excitement is about agents and automation. But Ramesh is clear: organizations that skip the data and governance groundwork in the "now" phase will find that their AI investments underdeliver—regardless of how sophisticated the technology is.

How should organizations think about build versus buy?

One of the most practical questions we asked was about the build versus buy decision—particularly relevant for organizations being pitched custom AI solutions when embedded options already exist in their ERP platform.

Ramesh shared the governing principle Infosys uses with clients:

Custom AI development is the second option—reserved for requirements that are genuinely specific to an organization's competitive differentiation and not covered by embedded solutions. And even then, Ramesh is disciplined about it:

• Value analysis upfront—what outcome does this custom build produce, and is it measurable?
• Architectural analysis upfront—how does this fit within the existing ERP and governance landscape?
• Five-year cost projection—including the often-overlooked cost of running custom AI (inference costs, model maintenance, token costs).

This is a framework that many organizations don't apply rigorously enough, particularly in the early enthusiasm of AI experimentation. The result, as Effie noted during the conversation, is pilot purgatory—AI investments that generate learning but not value. Ramesh's advice is to be intentional from the start about what success looks like and how you'll measure it.

What is the future of ERP?

We always end the podcast with the same question—and Ramesh's answer was worth quoting in full:

That phrase—intelligent backbone—captures something important. The organizations that will get the most from AI are not the ones that replace their operational foundations, but the ones that strengthen them and build on top. The data, process integrity, and governance that cloud ERP provides aren't constraints on AI ambition. They're what make AI ambition achievable.

Three things to take away from this conversation

1. Getting your data right is not optional. It's the single most common bottleneck Ramesh sees in AI engagements—and it's the work that has to happen in the 'now' phase, before agents can deliver meaningful value.
2. Start with embedded AI before building custom. SAP's portfolio of 300+ embedded AI solutions is the most practical starting point for most organizations. Custom development should be reserved for genuine competitive differentiation—and evaluated rigorously on cost versus value.
3. Governance isn't a back-office task. The AI governance layer Ramesh describes is what allows organizations to scale from 20 agents to enterprise-wide adoption without losing control of quality, compliance, or cost.

Listen to the full episode with Ramesh on The Future of ERP podcast.

FAQ