Starting small, creating big impact: The Smart Path to Autonomous ERP
Invoices can be approved automatically, purchase orders can be placed without human intervention and inventory can be replenished autonomously. Autonomous ERP is gaining momentum, and with growing interest comes the temptation to implement it immediately, particularly in areas where the potential benefits appear greatest.
That impulse is understandable, but it is not without risk. The areas with the greatest potential impact are often those where mistakes have the most serious consequences. If something goes wrong early in the process, trust in the system can disappear quickly. Confidence in the people who designed and implemented it may decline just as rapidly. A large-scale rollout is therefore not the best way to build trust in autonomous ERP. A measured approach is.
Step 1: Start small with processes where mistakes are manageable and easy to reverse.
Step 2: Set clear boundaries for what ERP can decide autonomously and when human intervention is required.
Step 3: Use pilots to learn, build trust, and scale step by step.
Step 1: Start small with processes where mistakes are manageable and easy to reverse.
Step 2: Set clear boundaries for what ERP can decide autonomously and when human intervention is required.
Step 3: Use pilots to learn, build trust, and scale step by step.
Een verstandig begin is een klein begin. Bepaal om te beginnen welke zich lenen voor automatisering, en bouw daar gerichte pilots omheen. Elke pilot, hoe bescheiden ook, levert waardevolle inzichten op. Zoals, de manier waarop het systeem tot een beslissing komt, waar de grenzen van de logica liggen en wat de organisatie nog nodig heeft om dit verder uit te delen.
Not all decisions involve the same level of risk. Some can be reversed easily, while others cannot. This distinction is crucial when selecting the first processes for autonomous ERP.
If a small reimbursement is approved incorrectly, the error can usually be corrected quickly. A strategic purchasing decision that leads to excess inventory, supply chain disruption or long-term contractual obligations is a different matter.
A second consideration is the reach of a potential error. Will its impact remain within the organisation, or will it affect external parties? While the system is still learning, errors should not have direct consequences for customers, suppliers or other stakeholders. Ordering the wrong office supplies is an internal inconvenience. Mishandling a customer order or failing to meet a delivery commitment can have much wider implications.
In practice, both considerations can be managed by setting clear thresholds. For example, the system may be authorised to make autonomous decisions about payments or purchase orders below a specified value, while anything above that value requires human approval. This principle is already widely used in organisations, as employees are rarely given unlimited approval authority. The same logic can be applied to autonomous systems, with the added benefit that thresholds can be adjusted as confidence in the system increases.
The purpose of the initial phase of an autonomous ERP implementation is not to prove that the technology works. It is to understand how it works.
The focus should be on gaining insight rather than delivering immediate results.
Record every decision the system makes, including the data it used and the reasoning behind the outcome. Then compare each decision with the choice a person would have made in the same situation. Where do differences arise? Can they be explained? Are they acceptable?
Discussing these questions with IT, operations and the people who work with the process every day will often provide more value than a long list of successful test results.
Each pilot also generates insights beyond this comparison. How does the system respond to exceptions? Where do organisational bottlenecks occur? Which decisions are not yet suitable for automation?
By tracking these findings consistently, organisations can build a reference framework that supports the implementation of more complex autonomous decisions in the future. Such a framework is equally valuable internally because it reveals knowledge gaps and governance issues that require attention. In this way, small pilots become the building blocks of an approach that can scale successfully.
This research was conducted as a quantitative online study commissioned by Dynamic People. The target group consisted of decision-makers responsible for IT, ERP and CRM solutions and/or financial decisions within organizations with 50 or more FTEs, operating in business services, wholesale and distribution, financial services or commodity trading. A total of 317 respondents participated in the study. The fieldwork was conducted by Markteffect between February 24 and March 9, 2026.