OpenEdge as the Foundation for Operational AI: Lessons from a Real‑World Coffee Buying Agent

AI is rapidly reshaping enterprise applications. But most enterprise AI projects don’t fail because of the models, they fail because they are disconnected from core operational systems like ERP.

For OpenEdge customers, the real question is: how do you bring AI into a production system without compromising trust, performance, or control?

A recent collaboration between inmydata and OpenEdge customers provides a clear, practical answer. As organizations move from experimentation to production AI, the real challenge is not intelligence—it is operational integration. By integrating an AI agent directly with an OpenEdge 12.8–based ERP system, this solution demonstrates how OpenEdge serves as the operational backbone for real, decision‑making AI—not experimentation, but production use.

Why AI Needs an Operational Core Like OpenEdge

Nick Finch, CTO at inmydata, explains that many AI initiatives fail not because of the models, but because operational data is hard to use correctly:

“One of the key difficulties we see people having in the AI space is joining that operational data to an AI workload.”

According to Finch, AI systems that operate on live business data must overcome three challenges:

• Speed – responses must be near‑instant, even when summarizing thousands of rows
• Security – agents must only access the data they are authorized to see
• Understanding – relational databases need context and structure for AI to work effectively

“If you just give your relational database to a language model, they’re not going to know what’s in the tables, what’s in the fields, and they’re not going to be very successful at building stuff out of that.”

This is where OpenEdge excels. The gap is now AI capability; it is context.

OpenEdge 12.8 as the System of Truth

In this implementation, OpenEdge remains the authoritative system of record. Operational ERP data — including inventory, demand, and manufacturing requirements — continues to live where it belongs: securely managed by OpenEdge 12.8.

Rather than replacing OpenEdge, the AI solution integrates with it using familiar OpenEdge technology, including PAS for OpenEdge. Finch describes the architecture clearly:

“We use app server technology to get the data from Progress. We have connectors into Progress applications using PAS for OpenEdge.”

Only a small OpenEdge procedure library is required to expose selected ERP data to the AI platform — no re‑platforming, no disruption.

This architecture helps make sure that OpenEdge remains the trusted operational core while AI extends its decision-making capability.

A Practical Use Case: AI‑Driven Coffee Purchasing

To make this tangible, the best way to understand the value is through a real business decision: coffee purchasing. To show what this enables, inmydata built an AI agent for a coffee‑roasting business — where purchasing decisions directly impact margins and supply continuity.

The AI agent combines:

• Operational data from OpenEdge ERP

• Stock levels

• Manufacturing demand

• Inventory availability

• External market intelligence

• Weather patterns in coffee‑growing regions

• Crop forecasts

• Market data

“We took a lot of operational data that they had in the business… and we’re now getting some of the expert knowledge that’s just sitting in the heads of their coffee buying experts and combining that with everything else.”

The result is not just better reporting; it is an AI agent that actively supports high‑impact procurement decisions. Grounded in real OpenEdge data, this “coffee buying expert” moves beyond generic assumptions to deliver actionable intelligence.

Capturing Expertise, Not Just Data

One of the most powerful aspects of this solution is how it extends beyond ERP data alone.

inmydata uses an AI‑driven interview platform to capture knowledge directly from experienced professionals. These interviews are transcribed, structured, and loaded into a RAG pipeline that complements operational data from OpenEdge.

Finch explains:

“There’s operational data — what’s actually happening day‑to‑day in your business — and then there’s the knowledge data. That’s all of your experts’ knowledge… What we’re trying to do is surface both of those to customers.”

This combination turns structured ERP data and human expertise into a single decision intelligence layer.

Making AI Accessible to Business Users

From the user’s perspective, the experience is intentionally simple. The AI agent is accessed through a conversational interface — not dashboards or reports.

Gary Lamb, CEO of inmydata, highlights why this matters:

“It works based on a chat interface, so it’s easy for any user to use… A CFO or a CEO doesn’t have to go to IT to get the information they need.”

Behind the scenes, OpenEdge continues to manage security, scale, and data integrity — allowing AI to feel approachable without sacrificing control.

What This Means for OpenEdge Customers

This coffee‑buying example illustrates a broader truth: AI adoption does not require replacing core systems; it requires extending them. OpenEdge is the enabler that makes operational AI practical, safe, and valuable.

With OpenEdge:

• AI solutions start with trusted operational data
• Performance and security expectations are preserved
• Customers can adopt AI incrementally — without rewriting core systems

As one of the interview participants noted, the fear many customers feel around AI often comes from not knowing where to start. OpenEdge provides that starting point.

OpenEdge and the Future of AI‑Driven Applications

As AI reshapes enterprise software, value is shifting away from complex user interfaces and toward intelligent services built on reliable operational platforms.

The coffee buying agent is just one example. The same OpenEdge‑driven pattern applies to forecasting, risk management, supply‑chain optimization, and beyond.

AI may be new — but trust, performance, and operational insight are not. The next generation of enterprise AI will not replace ERP systems; it will run through them. That’s why OpenEdge is not just part of the AI future — it is where operational AI becomes real.

Learn more about AI capabilities in OpenEdge.