Eradani Uses AI For New EDI And API Service
May 21, 2025 Alex Woodie
Eradani this week rolled out a new service aimed at helping customers to accelerate their EDI and API integration projects. What’s unique about its new Managed API and EDI Integration Service offerings is that it uses a large language model (LLM) to essentially take the role of the experienced integration specialist.
Eradani has been in the data integration business since Dan Magid founded the company in 2018. The Berkeley, California, company sells a tool called Eradani Connect that seeks to eliminate the complexity of data integration challenges involving the IBM i server, and to enable users to connect the IBM i to anything else that can work with APIs described using standard protocols, like REST, SOAP, and XML.
The shift into electronic data interchange (EDI) and managed services represents a bit of a shift for Eradani. Magid explains how the new offering emerged organically:
“We had Eradani Connect and we were growing very fast, getting customers who were using us to do integrations, doing APIs. But we had a lot of customers would say, well, I have this other integration that’s going on that I’m doing via EDI. Can you help me with that?” Magid explains to IT Jungle. “We look at it and say, well, we think we can. Let’s take a look. And we found out that EDI is actually pretty complicated, because it’s what we like to call the nonstandard standard, because everybody does it a little differently.”
EDI emerged in the late 1960s as a way for the US military to organize its supply chain. It was later adopted by companies in the retail, manufacturing, and transportation space. ANSI set the X12 standard in 1979, while the United Nations established EDIFACT to create a worldwide standard.
Despite the emergence of newer integration methods, like APIs and webservices that use REST, EDI continues to be widely used. Like the IBM i server itself, EDI has internal momentum that will not die.
Unfortunately for many EDI users on the IBM i platforms, many of the vendor that once existed to develop and support EDI mapping software have either died or been bought out by larger vendors that don’t invest in the software or the personnel needed to maintain it. The result is that many EDI projects are beset by long wait times and six-digit price tags. That’s an untenable situation for companies, but an opportunity for Eradani.
“Customers were telling us, well, when I need to onboard a new partner, I go to my EDI vendor and they tell me, okay, we think we can get you on the schedule in four to six weeks or in three months,” Magid says. “And then when we do that, then it’ll take us another week or two to actually get the integration done. And customers are saying that means I’ve now lost two months of business or a month of business with this potential customer.”
Rather than trying to attract sparse humans with EDI talent, Eradani did the next best thing: It trained an LLM to understand EDI. The company trained a Claude model from Anthropic to understand what’s going on with X12 EDI documents. As Magid explains, the hard part was training the model to understand what to do with the edge cases and the nuances.
“Everybody tweaks them a little bit. That’s one of the things we learned,” Magid says of the X12 documents. “So one of the things we did is we the first thing we did is we wrote a parser that will take an EDI document, take all the fields and turn it into JSON. So we now have instead of an EDI document, we’ve taken that EDI document and turned it into JSON. Then we have an AI-based mapper that says, okay, now I want to map this JSON data into my database. And that’s where the real kind of complexity of AI comes in.”
One might think that an EDI document is just mapping X12 field A to database column B. That would be incorrect. In fact, EDI documents are expressions of complicated business logic. That’s where the power of AI really shines, Magid says.
The Claude model doesn’t usually get the EDI perfect the first time through. There’s usually some back and forth with a human expert that’s required to get the new EDI documents hammered into shape. That’s where the managed services component comes in.
Instead of creating a shrink-wrapped software product to sell, like Eradani Connect, the company is offering AI-based EDI mapping as a managed service offering. Customers will be connected with an Eradani expert who knows how to use the software to get the right result. The customer is free to use the Eradani expert for as long as they need, and put them on as many projects as they require.
“Originally we’re selling a product. We’ll train you how to use it,” Magid says. “With the managed services, we’re saying, okay, we’re going to sell this to you, but we’re going to do it in a way that doesn’t require us to do a statement of work for everything we’re going to do, and do estimates and billing by the hour. We’re just going to do the subscription model. Say you have the subscription. You’re entitled to us doing one integration at a time for you, and we will just do that.”
The cost of Managed API and EDI Integration Service is $55,000 per year, in addition to the cost of the software. For more information, see www.eradani.com.
RELATED STORIES
Eradani Taps GenAI for IBM i Integration Boost
Eradani Debuts DevOps Suite for IBM i
Eradani Drives Simplification with API Tool Update
