Ep. 230 - AI in Manufacturing with Tom Hechtman of Sepasoft | Real ROI, MES on Ignition, Sepa IQ

Sep 25, 2025

Tom Hechtman, founder of Sepasoft and creator of Ignition MES modules, brings decades of manufacturing expertise. He delves into the real-world applications of AI on the plant floor, discussing the challenges of quality improvement, predictive maintenance, and the significance of reliable data. He explains the distinction between language-focused models and traditional machine learning, highlighting the importance of well-trained data for effective AI deployment. Tom also addresses the limited ROI of AI initiatives and shares insights on the future of AI in manufacturing, emphasizing continuous learning.

Ask episode

AI Snips

Chapters

Books

Transcript

Episode notes

ANECDOTE

From Integrator To MES Creator

Tom Hechtman built Sepasoft modules for the Ignition platform after running an integration business and working with Allen‑Bradley tech.
He created full MES suites including an ISA‑88 batch implementation running inside Ignition to serve regulated industries.

INSIGHT

AI Needs Outcome‑Driven Data

Manufacturing AI use depends on the target outcome: quality, scheduling, maintenance, or ERP integration.
AI needs relevant plant and business data to learn and cannot invent accurate operational facts on its own.

INSIGHT

LLMs Versus Machine Learning

LLMs handle language and must be connected to tools or agents to access sensor and production data.
Machine learning finds patterns in numeric features, while LLMs synthesize language and can call tools for real data.

Get the Snipd Podcast app to discover more snips from this episode

Get the app

Artificial intelligence is no longer just a buzzword in manufacturing. The pace of adoption has been incredible, yet the reality is far more complex than flashy headlines suggest. In this episode of Manufacturing Hub, Vlad Romanov and Dave Griffith welcome back Tom Hechtman, founder of Sepasoft, to explore how AI is actually being deployed on the plant floor, what barriers remain, and whether we are truly transforming manufacturing or simply tinkering at the edges.

Tom brings decades of experience building MES solutions for manufacturers around the globe. From his early days in the Midwest working with Rockwell Automation technology to launching Sepasoft’s Ignition MES modules and now leading the development of Sepa IQ, Tom has been at the forefront of data, analytics, and system integration. His insights bridge the gap between hype and practice, helping us understand where AI creates real ROI, where it still falls short, and how to build the foundations for success.

Throughout the conversation we dive into the challenges of quality improvement, predictive maintenance, scheduling optimization, and contextualizing plant data. We discuss the importance of trust in both data and AI-generated outputs, the economics of running LLMs and machine learning models, and why cybersecurity and data governance cannot be an afterthought. Drawing on the recent MIT study that revealed only 5 percent of AI projects make a measurable P&L impact, Tom helps us unpack what manufacturers need to do differently if they want to avoid being part of the 95 percent that fail.

We also get an update on Sepa IQ and how customers are using it to connect plant floor data, structure it for AI and analytics, and prepare for advanced scheduling and predictive tools. From lessons learned working with early adopters to practical advice on starting small, Tom makes it clear that manufacturing AI is a journey that requires technical expertise, domain knowledge, and cultural change.

Whether you are an executive evaluating AI investments, a controls engineer curious about new tools, or a plant manager wondering how to get real results, this episode delivers a balanced, practical, and in-depth perspective on the future of AI in manufacturing.

Timestamps
00:00 Introduction and AI in every manufacturing conversation
03:00 Tom Hechtman background and the origins of Sepasoft
05:00 MES modules, batch processing, and the evolution of Sepa IQ
08:00 Defining manufacturing AI and the role of plant floor data
13:00 Quality improvement and predictive analytics opportunities
20:00 Foundational challenges with legacy systems and data collection
24:00 Insights from the MIT study on AI adoption and ROI
32:00 Training data, context windows, and the economics of LLMs
44:00 Sepa IQ customer feedback and scheduling optimization
50:00 Trust, hallucinations, and cybersecurity considerations
59:00 ICC announcements, demos, and proof of concept program
1:03:00 Predictions for AI in manufacturing and career advice

References mentioned in the episode
MIT study on AI in business: https://mitsloan.mit.edu/ideas-made-to-matter/how-to-get-ai-to-pay-off
Sepasoft Blog on AI and manufacturing: https://sepasoft.com/blog/
The One Thing by Gary Keller: https://www.the1thing.com/

About the guest
Tom Hechtman is the founder of Sepasoft, a leading provider of MES modules for the Ignition platform by Inductive Automation. With decades of experience in manufacturing, integration, and software development, Tom has helped companies worldwide improve efficiency, quality, and data visibility. Learn more at https://sepasoft.com/
and connect with Tom on LinkedIn: https://www.linkedin.com/in/tom-hechtman-32b6b66/

About the hosts
Vlad Romanov is the founder of Joltek, a consulting and integration firm focused on helping manufacturers modernize systems, bridge IT and OT, and drive digital transformation. Learn more at https://www.joltek.com/
and connect on LinkedIn: https://www.linkedin.com/in/vladromanov/

Dave Griffith is an industrial automation consultant and co-host of Manufacturing Hub. He works with manufacturers and integrators to identify opportunities, manage digital projects, and build future-ready operations. Connect with Dave on LinkedIn: https://www.linkedin.com/in/davegriffith/

If you enjoyed this conversation, make sure to subscribe to Manufacturing Hub for more weekly discussions with industry leaders about automation, digital transformation, and the evolving world of manufacturing.