ERP Selection in 2027: Why AI Agents Are Changing the Questions Companies Need to Ask
For years, selecting an ERP system followed a fairly predictable process.
Define the requirements. Evaluate the vendors. Sit through demonstrations. Compare implementation plans. Negotiate the contract. Choose a system.
That process still matters. But heading into 2027, there is a new question sitting in the middle of almost every serious ERP evaluation:
What happens when the ERP system starts doing the work instead of simply helping people do it?
That is no longer a theoretical question.
Microsoft, SAP, Oracle and other major enterprise software vendors are moving quickly from traditional automation and AI copilots toward AI agents that can work across finance, supply chain, procurement and other business processes.
Microsoft describes its direction for Dynamics 365 as “agentic ERP,” where people and AI work together to make decisions and execute processes. Its recent Dynamics 365 updates include advances in areas such as account reconciliation and accounts payable automation. Build the future of agentic ERP with new Microsoft Dynamics 365 capabilities [microsoft.com]
SAP is making a similar move with Joule Agents and Assistants that can execute multistep tasks across SAP and non-SAP systems, with governance and human oversight built into the environment. Joule Agents and Joule Assistants [sap.com]
Oracle has gone even further in its terminology, introducing Fusion Agentic Applications designed to reason, make decisions and execute work across finance, supply chain, HR and customer-facing processes. Oracle AI for Fusion Applications [oracle.com]
For companies considering an ERP replacement, this changes the selection conversation significantly.
ERP Is Moving From a System of Record to a System of Action
Traditional ERP systems are very good at recording what happened.
A purchase order was created. Material was received. An invoice was posted. Inventory moved. A production order closed.
Modern ERP systems have steadily become better at telling us why something happened and what might happen next.
AI agents introduce another step.
The system may increasingly be able to determine what needs to happen next and initiate the work itself.
Microsoft describes this evolution as moving business applications from systems of record toward systems of action, with AI agents operating using the business rules, permissions and audit trails already established within Dynamics 365. Microsoft Dynamics 365: An agent-ready business applications platform [microsoft.com]
That distinction matters enormously for ERP selection.
When an ERP system simply gives someone a recommendation, an employee still owns the final action.
When an agent can execute a transaction, contact a supplier, resolve an exception or move a financial process forward, the stakes become higher.
Now ERP selection involves more than functionality.
It involves governance, data quality, security, process ownership and decision authority.
Manufacturers Should Pay Particular Attention
For manufacturers, this shift could be substantial.
Think about the amount of ERP activity generated by a typical manufacturing operation:
- purchase requisitions
- production orders
- supplier exceptions
- inventory shortages
- quality issues
- demand changes
- scheduling conflicts
- invoice discrepancies
- late shipments
- material availability problems
Historically, ERP systems have surfaced many of these issues and waited for someone to respond.
The emerging model is different.
An AI-enabled ERP may identify an exception, evaluate the available information, recommend a response and, within established permissions, take action.
Oracle, for example, is already positioning agentic applications around areas such as inventory planning, production readiness, manufacturing and accounting operations. 26C roadmaps—agents, agentic apps, AI Studio Skill [blogs.oracle.com]
SAP is similarly positioning Joule Agents around business processes that span finance, supply chain, HR, procurement and customer experience. SAP’s AI Vision Becomes Reality at SAP Connect 2026 [sap.com]
This is why we believe manufacturers entering an ERP selection today should consider their future AI operating model even if they have no intention of turning on autonomous functionality immediately.
You are not selecting an ERP solely for next year.
You may be selecting the core operating platform your business will use for the next decade.
The ERP Demo Needs to Change
ERP demonstrations have traditionally focused on functionality.
Can the system configure this product?
Can it schedule this job?
Can it handle this costing method?
Can it process this type of order?
Those remain important questions.
But companies should now add another layer to their demonstration scripts.
Ask the vendor what the AI can actually do.
There is an important difference between:
AI that summarizes information
and
AI that recommends an action
and
AI that executes an action.
Those three capabilities should not be lumped together under an “AI” checkbox on an RFP.
Ask vendors to demonstrate specific scenarios.
For example:
A critical raw material will run short three days before a production order is scheduled. What does the system do?
Does it merely identify the shortage?
Does it recommend suppliers?
Does it create a purchase requisition?
Can it contact an approved supplier?
Can it change the production schedule?
And perhaps most importantly:
Who has to approve each step?
That conversation tells you much more than asking whether the ERP vendor “has AI.”
Data Quality Is Becoming an ERP Selection Issue Again
There is an uncomfortable reality behind all of this.
AI cannot magically repair bad operating discipline.
If your item masters are unreliable, bills of material are inconsistent, routings are outdated and supplier information is incomplete, adding sophisticated AI doesn’t eliminate the underlying problem.
It may simply allow bad information to travel faster.
That makes data readiness an increasingly important component of ERP selection and implementation.
Companies should evaluate:
- Item master accuracy
- Customer and supplier data
- Bills of material
- Routings
- Inventory accuracy
- Chart of accounts
- Historical transactional data
- Security roles and permissions
- Approval structures
The cleaner the operational foundation, the more useful AI-enabled ERP capabilities can become.
This is also why ERP implementation should not be viewed strictly as an IT project.
ERP implementation remains a business transformation project involving processes, people, controls and data.
AI makes that more important, not less.
AI Is Also Changing ERP Pricing
There is another issue receiving considerably less attention than the technology itself: how companies will pay for all of this.
Traditional enterprise software economics have largely revolved around users, modules and subscriptions.
Agentic AI complicates that model.
AI usage generates variable computing costs, and software vendors are experimenting with usage-based, outcome-based, credit-based and hybrid pricing approaches.
How SaaS Pricing Must Change for Agentic AI in 2026 reports that 57% of surveyed hypergrowth SaaS companies had already adopted hybrid pricing models, illustrating how quickly software economics are changing. [bcg.com]
ERP buyers therefore need to think beyond today’s subscription quote.
When evaluating an ERP contract, ask:
- Is AI functionality included in the base subscription?
- Which AI capabilities require additional licensing?
- Is AI usage metered?
- Are there transaction or consumption limits?
- How will pricing change as AI usage increases?
- Are future agent capabilities automatically included?
- Can pricing change when the vendor changes its AI packaging?
- What happens at renewal?
These questions belong in ERP contract negotiations today.
Waiting until renewal may be too late.
Don’t Buy ERP Based on an AI Demo
There is also a danger of overcorrecting.
AI should not suddenly become the deciding factor in every ERP selection.
A polished AI demonstration cannot compensate for weakness in:
- manufacturing functionality
- financial controls
- inventory management
- planning
- reporting
- integrations
- implementation capability
- support
- industry fit
ERP buyers have seen waves of technology hype before.
Cloud ERP was supposed to solve everything.
Digital transformation was supposed to solve everything.
Machine learning was supposed to solve everything.
Now AI agents risk becoming the next checkbox companies chase without first defining the business problem.
The right approach is much simpler:
Start with the business.
Where does the organization lose time?
Where are decisions delayed?
Where is information unreliable?
Where do employees perform repetitive work?
Where are errors expensive?
Where do customers feel the impact?
Then determine whether ERP functionality, process redesign, automation or AI can improve the outcome.
Technology should follow the business case.
Not the other way around.
Five Questions We’d Add to Every ERP Selection in 2027
If we were beginning an ERP evaluation today, these questions would be included in the process.
1. What Can Your AI Actually Execute?
Ask vendors to distinguish between summarization, recommendation and autonomous execution.
Demand demonstrations using your actual business scenarios.
2. How Is AI Governed?
Understand approval thresholds, security permissions, audit trails and escalation procedures.
If the system can take action, you need to understand precisely where its authority begins and ends.
3. What Data Does the AI Use?
Determine which ERP records, external sources and historical data feed the AI.
More importantly, determine whether your organization actually has the data quality necessary to support it.
4. How Will We Pay for AI?
Understand today’s pricing and the contract language governing tomorrow’s pricing.
This is rapidly becoming part of total ERP cost of ownership.
5. What Happens If We Don’t Use the AI?
This may be the most overlooked question.
The ERP still needs to be the right ERP.
A manufacturer should not select the wrong core operating platform simply because one vendor currently has the most impressive AI demonstration.
The Role of an Independent ERP Advisor Is Changing Too
Independence has always mattered in ERP selection.
It becomes even more important when vendor roadmaps are moving this quickly.
Companies need someone at the table asking questions that do not begin with:
“What does the software vendor want to sell us?”
The better questions are:
“What does the business actually need?”
“What problem are we solving?”
“What risks are we accepting?”
“What are we committing to contractually?”
“Will this still make sense five years from now?”
That includes evaluating functionality and implementation capabilities, but increasingly it also means evaluating AI governance, data readiness, future licensing models and the vendor’s broader technology direction.
ERP Selection Has Become a Future-State Decision
The ERP market entering 2027 looks different than it did even two years ago.
Microsoft is openly building toward agentic ERP. Build the future of agentic ERP with new Microsoft Dynamics 365 capabilities [microsoft.com]
SAP is deploying Joule Agents capable of performing multistep business processes. Joule Agents and Joule Assistants [sap.com]
Oracle says its Fusion Agentic Applications can reason, make decisions and execute complex processes across major enterprise functions. Oracle AI for Fusion Applications [oracle.com]
This isn’t a reason to rush into AI.
It is a reason to become more disciplined about ERP selection.
Companies should select systems based on where their operations are going, not simply where they are today.
The winners won’t necessarily be the companies with the most AI.
They will be the companies that combine a strong ERP foundation, reliable data, disciplined processes, sensible governance and AI capabilities that solve real operational problems.
That is a much higher bar than checking an AI box on an RFP.
And frankly, it should be.
About Impact North
Impact North, Inc. provides independent ERP advisory services focused on ERP software selection, ERP contract negotiation, and ERP implementation. We work with manufacturers, distributors and executive teams to reduce ERP risk, evaluate technology objectively and build systems and processes that support long-term growth.
