Lean, composable, and agile: How ERP is evolving in the agentic AI era

Is enterprise resource planning still necessary in the age of AI? 

Traditional enterprise resource planning is stable and trusted, but it can be rigid and costly. As agentic AI matures to handle greater levels of autonomy and complexity, there is a question about whether ERP is still needed.

It’s a compelling idea: doing away with inflexible and expensive platforms and replacing them with AI solutions. The truth, however, is that ERP is the language and fabric of the enterprise, enabling critical capabilities such as auditability and semantic and data context for AI at scale. What’s more, AI consumes and computes data; it is not the data itself. ERP will persist, but it should evolve. A leaner, more agile vision for ERP is necessary for the agentic era.

ERP modernization on the rise—and evolving toward agentic ERP

Marketplace trends indicate enterprise investment in ERP is ticking up. A 2025 Deloitte study found that of organizations surveyed, 43% are investing in ERP, up from 35% in 2024.¹ Gartner forecast that spending on public cloud services would reach $1.48 trillion by 2029², suggesting ongoing ERP migrations to cloud platforms.

Companies are investing to change, and ERP is moving away from unwieldly platforms that are ill-suited to dynamic enterprise environments and toward a lighter-weight, AI-driven, modular approach—an autonomous ERP ecosystem or “agentic ERP.” As the lines between IT and business domains fade, there is an opportunity to transform how the organization accesses data and insights, and how innovation can be achieved with agents and the reimagination of work.

Benefits of agentic ERP: Breaking constraints to foster innovation

Enterprises rely on ERP to provide all parts of an organization with access to company data through standardized, rules-based processes. But as business units independently adopt AI solutions for their specific needs, the resulting tech-debt-heavy ERP can inhibit agentic experimentation, innovation, and end-to-end workflow transformation.

Agentic ERP helps address traditional limitations through three principles:

By decoupling enterprise data and insights from a less-flexible user interface, organizations can pursue AI agent adoption and the transformation it entails.

How does agentic ERP work?

Agentic ERP keeps the core lean and composable with rules, a database structure, and native workflows that are rigid where needed, such as in financial accounting, compliance, accounts payable, and other processes with tight controls and governance.

Meanwhile, it boosts upside potential in the surrounding application ecosystem through insight-sharing, agentic enablement, and digital empowerment. A composable app layer lets agents operate peripherally according to core rules while leveraging low-code UIs, an analytics platform, and application programming interfaces (APIs) to access enterprise data and insights.

With this approach, agents effectively become the data and insights interface: They connect to back-end systems and allow end users to fit applications to their needs and innovate with greater flexibility.

Examples of autonomous ERP innovation

Taking the next steps toward AI agents for ERP

Now is the moment to begin to change. There is an imperative for many enterprises to reimagine how they implement and architect ERP. There is also an opportunity to better harness data more powerfully in the core while gaining the flexibility to adopt agents.

Agentic ERP is a bold vision, and change takes time. Rather than a wholesale replacement, we are likely to see a phased evolution as organizations invest in a modular, API-driven ERP ecosystem. Many enterprises may not be ready to grab the ERP core with both hands and upend the general ledger and financial accounting—these semantic layers must be rigid for regulatory and audit reasons, and structures and standards are important. Even as a vector database enables AI applications, a relational database is still necessary as it is the underlying structure for AI-enabled ERP solutions.

The stages of agentic ERP modernization are likely to begin with changes to UI/UX, followed by modularization and effective API integration with other value-added systems and platforms.

Key questions for discussion

There will be key questions and barriers to address as organizations transition from transaction entry to transaction orchestration:

• Considering full-time equivalent (FTE) efficiencies and growth enablers associated with ERP, how does the broader ecosystem come together to create value for the organization?
• How do we establish universal agreement on new document and process formats?
• Is technological readiness sufficient to scale agentic ERP to large enterprise transaction volumes?
• Critical for auditability and regulatory hurdles: Will auditors trust AI-generated financials?
  
  

To the latter question: Trust will grow out of engagement, and audit and assurance teams must be involved in the discussions about ERP evolution—particularly as it relates to confidence in AI-generated financial results and compliance implications.

Agentic AI for ERP in summary

Bringing together a structured core with an ecosystem of agents and applications unlocks insights and value while maintaining a lean, composable ERP. With AI agents for ERP, enterprises have an opportunity to access agentic AI capabilities while enabling an agile business that can grow, acquire, expand, and scale.

Endnotes

1. Tim Smith et al., “AI is capturing the digital dollar. What’s left for the rest of the tech estate?,” Deloitte, October 16, 2025.
2. Gartner, “Forecast: Public cloud services, worldwide, 2023-2029, 3Q25 update,” September 26, 2025.