Deploying AI with ERP reminds me of when I first started implementing ERP for small and mis-sized companies, I had many clients that did not understand why they needed a consultant. Why doesn’t the software just work? I would spend a lot of time explaining that complexity of the software, configuration decisions, re-engineering processes, data migration, etc. Now that we are adding AI to their solutions, I hear a lot of the same questions, why doesn’t it just work?
If you want to leverage generative AI, like ChatGPT or Claude, it does. You can interact with AI through their interface to ask questions and generate content. Sometimes this capability is even embedded in the software. For example, Business Central uses AI in the item master to help generate Item Descriptions based on the item attributes. Meeting notetakers are another example of AI that works without too much effort.
But to really leverage AI, it takes a more thoughtful approach, and you may need additional expertise or capacity. You may want a consultant to fill the following roles:
- Solution Architect + Business Translator
- Bridge between AI developers and business process owners
- Define goals, map pain points, identify automation opportunities
- Assist with tool or service selection
- Data + Process Analysis
- Understand what and where data lives in the ERP
- Help structure inputs for training or guiding agents
- Implementation + Change Management
- Guide ERP users through pilot and rollout
- Validate outcomes, manage exceptions, build trust in the system
- Governance + Risk Management
- Ensure regulatory, privacy, and ethical standards are met
- Design review checkpoints and escalation paths
- OpenAI Functions / Retrieval-Augmented Generation (RAG)
- Integrating agents with business tools and data
- Provide training to AI where necessary
- ERP Integrations via APIs
- Connect AI solutions to the ERP using the connectors available for that software
- Build connector when the appropriate connector does not exist
In addition to architecting, developing and deploying solutions, it is also important to understand the total cost of ownership. AI is not free, in fact, it can be extremely expensive when you evaluate all the cost components. Just like process automation, the cost/benefit should be fully evaluated. An expert can provide insight into cost components you haven’t considered.
You may not require assistance in all these areas, but most organizations will have gaps, especially if they have not already deployed AI solutions. Achieving the desired outcome while mitigating risk, managing costs, and maintaining security will require comprehensive knowledge or your business and the available technology.
In short, using AI with your ERP requires thoughtful integration and management—consultants help ensure it aligns with business needs and works effectively.
You can learn more about AI in Gary Feldman’s Posts How is the Question at Microsoft AI Days or The Rise of AI Agents in ERP