Background Vector
Background Vector

Bringing AI Into the Workflow: Har-Conn’s Oven Calibration Process

I processor chip with circuit board connections overlaid on city skyscrapers, representing artificial intelligence in business technology

At The Support Group, we have deep expertise in Claris FileMaker—and it’s a platform we continue to value and support. But what truly drives our work is understanding your business and solving the challenges behind it. In many cases, FileMaker is the right fit. And when it’s not, we help guide you toward the best solution for your needs.

Meeting Har-Conn Where They Were

Har-Conn Chrome Co. is a premier special process firm founded in 1948.  The company provides high-technology electroplating and coating services to clients in aerospace, medical, automotive, and electronics industries. Their work is governed by strict quality standards, and the documentation to support those standards has to be accurate and consistent every single time.

We’ve been working with Har-Conn on a broader FileMaker solution to help manage their operations.  However, along the way we identified a specific pain point that needed its own answer: their oven calibration process was almost entirely manual. This was taking up far more time than it should. Rather than force it into the FileMaker project, we asked what the best tool for that particular job actually was.

The Problem: Too Much Manual Work

Oven calibration at Har-Conn involves running an industrial oven through up to three target temperatures while a 10-probe recording unit captures temperature readings every two minutes. On the surface, that sounds straightforward. In practice, what followed was a lot of careful, time-consuming manual work.

Before a technician could evaluate a single reading, they first had to consult two separate certification PDFs.  The first one documenting the offset values for each individual probe, and another for the thermocouple wire present in the oven during testing. Because these offsets vary by temperature, the procedure required manually identifying and pulling the right values for each target temperature, then applying those corrections by hand to every raw data point exported from the recorder.

Only after all of that could the real analysis begin.  First, scanning through the corrected data to find the first window of 15 consecutive readings where all 10 probes fell within the acceptable range — at each target temperature. If the oven passed, the technician then compiled a report summarizing the overall result, the qualifying time window, and the min/max corrected temperatures per probe.

It was meticulous work.  Because so much of it was manual, it carried real risk of error in a context where accuracy matters a great deal.

The Solution: The Right Tool for the Job

When we sat down with Har-Conn to understand the problem, it became clear pretty quickly that this wasn’t a software problem. It was a document processing and data analysis problem — and that’s where AI tends to shine.

So that’s the direction we went. Part of our process involved testing multiple AI platforms to evaluate how well each one could interpret and extract structured data from certification PDFs, apply that data accurately to a raw dataset, and work through the pass/fail logic reliably. After comparing results, we settled on Claude as our tool of choice.  We selected it because of its accuracy, consistency, and ability to follow complex, multi-step instructions.

The result is a solution that takes three inputs — the raw CSV from the temperature recorder and the two calibration PDFs — and handles everything else automatically. It extracts the correct probe and wire offsets at each target temperature, applies them to the full dataset, identifies the first qualifying 15-consecutive-reading window, and produces a clean, formatted report. What previously took significant time and careful manual effort now takes minutes.

What This Really Means

This project was about understanding a client’s workflow well enough to recognize where the real bottleneck was.  From there, we chose the tool that made the most sense. In this case, that happened to be AI — but it could just as easily have been something else.

At TSG, we think of ourselves as business consultants first and technology experts second. Our goal on every engagement is to understand what you’re actually trying to accomplish and applying the right tools for the job. What doesn’t change is our commitment to clear communication, thoughtful delivery, and solutions that genuinely make your work easier.

What TSG Can Do For You

The Support Group works with organizations across manufacturing, aerospace, healthcare, financial services, higher education, nonprofits, and more. If you’ve got a process that’s taking longer than it should, or a problem you haven’t quite figured out how to solve, reach out to us. Let’s figure it out together!

Related Articles

Team collaborating on business reporting using a flexible FileMaker application
Business requirements change. Learn why FileMaker makes it easy to adapt reports, workflows, and databases without lengthy development cycles.
Illustration of AI connecting FileMaker 2026 with multiple business systems using intelligent data workflows.

Last month, we explored the most impactful developer-focused features in FileMaker 2026: from Remote Backup and Standby Server to native PDF generation and other enhancements that make building and maintaining applications easier. But there was one topic we didn’t touch

FileMaker 2026 new features highlighted as gifts under a Christmas tree

Do you feel that? A little magic in the air? It’s been feeling like Christmas in June around here for FileMaker programmers. Claris has released a new version of Claris FileMaker, and they’ve stuffed plenty of goodies into developers’ stockings.