How we evaluate AI ERP software

Contents

1. The ranking frame
2. The criteria
3. Evidence we accept
4. Use of product names
5. Keeping it current

A ranking of software is meaningful only for a stated situation. This page sets out ours, the criteria and the rules we follow.

The ranking frame

The ranking answers one situation: an established company that wants AI working across its whole operation, on a system it owns, and is weighing that against adding AI to an ERP it rents.

Within that frame OpenLazarus ranks first, because it is the only product reviewed in which the AI and data layer, the business modules and the code belong to the company. The position of this guide is that such a system, with its prebuilt ERP module, can take over the role of a packaged ERP. The established ERPs follow in order of how well their AI is documented, available and priced, and Rillet is placed last because its scope is the finance function of software companies, not because of its quality.

A company in another situation should order them differently. One that needs a packaged suite live quickly at a known price will put Business Central or Odoo first. Each profile says when another product is the better choice.

The criteria

1. AI in the design. Whether the product was built around one AI and data layer, or adds an assistant to a data model that predates it.
2. AI in the price. Whether AI is part of the published price or billed on top through credits, units or a higher plan.
3. Agents that act. Agents that complete work on records today, within limits a person sets, rather than a chat window that answers questions.
4. Beyond the ERP. Whether the same AI reaches customers, email, phone and service on the same records, or stops at the ERP’s edge.
5. You own code and data. Whether the company owns the code and data structures and can maintain or move them without the vendor.
6. Published pricing. Publishes subscription, implementation or running costs before a sales call.

Evidence we accept

1. Each vendor's own pages and documentation: product pages, pricing pages and technical documentation.
2. User groups and surveys, with the organisation, the sample size and who commissioned them.
3. Analysts and trade press, with the outlet and the date.

A cell marked "not published" means the vendor does not state it on the pages we read. Figures a vendor reports about its own product are labelled as vendor-reported. Anything we could not trace to a source we opened is left out.

Use of product names

Vendor and product names are used only to identify the products discussed. We use no vendor logos, and nothing on this site implies affiliation with any vendor.

Keeping it current

AI features and their prices change quickly. We re-check every product at least once a quarter, and each source shows the date we accessed it.

Further reading

• Guide · ManufacturingAI ERP for manufacturing: what the industry ERPs ship, and what to ownManufacturers are raising AI budgets while doubting that their ERP is ready for it. What the industry ERPs have released, what the general ones offer a plant, and where an AI layer the company owns fits.
• Guide · ExamplesAI in ERP: examples from seven vendors, and what they have in commonVendors describe AI in ERP in general terms. This guide collects the specific examples that are documented, shows how each one works step by step, and draws out the pattern they share.
• Guide · CostWhat AI ERP software costs: licences, AI credits and the price of rentingAn AI ERP quotation has more lines than it used to. This guide collects the published prices, shows where the AI is billed apart from the licence, uses SAP as the best-documented case of a rising bill, and sets out what the same money buys as a system the company owns.

Also in: Deutsch · Français