)
)
Finance leaders evaluating AI-enabled ERP platforms should recognize that vendors are at very different stages of AI maturity.
While many platforms now advertise “AI-native” capabilities, the underlying architectures often reflect earlier stages of the evolution described above.
The following landscape illustrates how leading ERP vendors appear to be progressing along the Connected → Generative → Agentic → Cognitive continuum.
Vendor | Current Stage | Observations |
|---|---|---|
Oracle Fusion Cloud ERP | Agentic | Embedded assistants and AI agents automate tasks but rely heavily on external model inference and AI services |
SAP S/4HANA | Agentic | Joule copilot and AI agents orchestrate workflows using generative AI and enterprise data context |
Workday | Generative → Agentic | Strong analytics and AI assistance with emerging agentic automation initiatives |
Microsoft Dynamics 365 Finance | Generative → Agentic | Copilot-driven assistants expanding into workflow automation |
NetSuite | Generative | AI-powered analytics and generative capabilities primarily focused on insight generation |
Sage Intacct | Generative | AI used primarily for forecasting, anomaly detection, and analytics assistance |
Infor | Generative | Embedded analytics and ML capabilities with growing automation features |
Rillet | Agentic | AI-first accounting workflows with heavy automation and external AI inference |
DualEntry | Agentic | AI-driven ERP challenger automating finance workflows through agent-based systems |
Campfire | Agentic → Cognitive | Emphasis on domain-specific accounting models and AI-native architecture |
Light | Agentic | AI automation across finance workflows with external model dependency |
Large ERP Platforms: Moving Toward Agentic Automation
The largest enterprise ERP vendors are currently transitioning from generative assistance toward agentic automation.
Platforms such as Oracle Fusion and SAP S/4HANA have introduced AI copilots and embedded agents designed to automate workflows and support decision-making directly inside the ERP environment.
For example:
Oracle Fusion now offers embedded AI agents designed to automate business processes and provide contextual assistance within the application environment.
SAP’s Joule assistant enables conversational access to business processes and can coordinate multi-step workflows using enterprise data context.
These developments represent meaningful progress in AI adoption.
However, in most implementations the core reasoning capabilities still depend on external model services, meaning operational automation remains tied to external AI inference.
Mid-Market ERP Platforms: Primarily Generative
Most mid-market ERP systems remain primarily in the generative stage.
Platforms such as NetSuite, Sage Intacct, and Infor have focused their early AI investments on:
Predictive analytics
Anomaly detection
Forecasting assistance
Natural language reporting
These capabilities behave much like an analyst embedded within the system, helping interpret financial data but rarely executing operational workflows autonomously.
Agentic automation is beginning to appear on vendor roadmaps, but large-scale workflow automation is still emerging.
AI-Native Finance Platforms: Automation First
A newer category of finance systems has emerged with automation as a core design principle.
Platforms such as Rillet, DualEntry, Campfire, and Light are often described as AI-native finance platforms, emphasizing automation of accounting workflows and operational tasks.
For example, AI startup DualEntry positions itself as an AI-driven ERP alternative designed to automate financial workflows and simplify migrations from legacy systems.
These platforms typically fall within the agentic stage, where AI systems actively execute workflows across financial operations.
However, most still rely heavily on external generative models for reasoning, meaning their automation remains dependent on the economics and availability of external AI services.
The Gap Between Agentic and Cognitive Systems
The distinction between agentic and cognitive systems is subtle but important.
Agentic systems can automate workflows.Cognitive systems can reason about financial activity internally.
In agentic architectures:
AI agents orchestrate workflows
Reasoning occurs in external models
Operational capacity scales with token consumption
In cognitive architectures:
Domain intelligence is embedded within the platform
Deterministic services manage financial logic
External models are used selectively rather than continuously
This distinction has significant implications for:
Cost predictability
Operational reliability
Governance and auditability
Why This Matters for CFOs
From a finance leadership perspective, the maturity of an AI-enabled ERP system should not be measured solely by the presence of AI features.
Instead, leaders should ask a more fundamental question:
Does the system rely on AI to help people operate it, or can the system itself reason about financial operations?
That difference determines whether AI becomes:
A layer of automation sitting on top of ERPor• an architectural capability embedded within the financial platform
The organizations that recognize this distinction early will be better positioned to design finance systems that deliver both automation and economic predictability as AI adoption accelerates.
)
)
)
){kind=icon}
