Standing for PSI in a Steel City

Representing PSI at the ISA Kolkata Section Technical Symposium in Bokaro Steel City, a talk on the PSI Chatbot, and a room full of people who actually run steel plants.

Event details

Start date and time: 15 May 2026 at 3:30 am
End date and time: 16 May 2026 at 11:30 am
Host: ISA-Kolkata Chapter, SAIL, Bokaro
Place: Bokaro Steel City, Jharkhand, India
Published: 19 July 2026 at 6:23 am

Some of the work we do never really gets seen. You solve the problem, close the ticket, and move on. So when you get a chance to stand up in front of a room and actually talk about it, to say this is what we've built and this is why it matters, it means something. That's how I felt representing PSI Software SE at the ISA Kolkata Chapter event in Bokaro Steel City.

Bokaro is the right kind of place for a conversation like this. It's a steel town through and through, built around furnaces and shifts and the endless movement of material. The people who show up to an ISA gathering here aren't spectators. They run plants. They know what a bad night in production feels like, and they know their way around a log file. Talking to that audience about Industrial AI isn't about dazzling anyone. It's about earning a nod from people who have seen a lot of technology come and go.

I'll be honest, I didn't want to open with the technology. I wanted to open with the pain, because that's the part everyone in the room already knows in their bones.

Taking the stage in Session I — “The Future of Industrial Automation with the advent of AI, ML & Analytics.” - ISA-Kolkata

The quiet weight of too much data

Anyone who has worked on enterprise software for a steel plant understands the quiet weight of too much data. Log files that run into gigabytes. Documentation you could get lost in for a week. Support teams that are thin, stretched, and quietly bleeding knowledge every time an experienced colleague retires or moves on. The frustrating part is usually not that the answer doesn't exist somewhere. It's that finding it takes time you don't have, and a kind of instinct for where to even start looking.

I've watched sharp, experienced people burn an afternoon chasing a single issue. Or trying to remember how something similar got fixed two years ago. That struggle is real, and I think naming it plainly matters more than any slick demo. For me, Industrial AI has never been about the technology being clever. It's about handing those people back their time. When I put it that way, I could feel the room settle in, because I was describing their Tuesday, not some far-off future.

It's about handing those people back their time. I was describing their Tuesday, not some far-off future.

So that's where our chatbot comes in

What we've done is embed an AI assistant into our PSImetals platform. You ask it a question in plain language, the way you'd ask a colleague who happens to know everything, and it goes and pulls the answer from the factory model and the documentation. No query to write. No syntax to memorise. You just ask.

Now, the moment you talk about AI to an industrial crowd, one thing is on everyone's mind, even if nobody says it out loud. Can I trust it? Everybody's heard the stories about AI that confidently makes things up. And in our world, a confident wrong answer is worse than no answer at all. Nobody wants a co-pilot that hallucinates over a production database.

So I spent a bit of time on how we avoid exactly that. The trick is that the assistant isn't guessing. It's connected to the real data through something called the Model Context Protocol. I usually describe it as a kind of universal connector, a secure bridge between the AI and the databases, files, and production history it needs to see. Because it's answering from what's actually there, you get the real answer instead of an invented one. And work that used to eat up hours starts taking minutes. Sometimes seconds.

That's a big claim, and it's only worth making if you've actually done the engineering to stand behind it. We have. Being able to say that on PSI's behalf, and mean it, is a very different thing from just selling a story.

The best conversations happened at the booth — after the slides were done. - ISA-Kolkata

Here's the part I find harder to put into words

Why it made me proud. I've been with PSI for more than thirteen years now. That's a long time to stay anywhere, and it's not something I say lightly. Over those years I've watched the company grow, watched our software end up inside some of the most demanding metals plants in the world, and seen just how much genuine care goes into the things we ship. To stand on an industry platform and represent all of that, to be the voice of it even for one afternoon, got to me a little.

Part of the pride was in the substance. I wasn't up there selling smoke. I was showing something we've actually built, something that actually works. Part of it was the audience. Serious people, from a serious industry, who don't hand out their attention for free. Earning their questions felt worth more than any easy applause. And part of it, honestly, was just belonging. Being able to say “we” and mean a company that has trusted me and grown alongside me for over a decade.

There's another kind of pride too, one I didn't fully expect. Steel and artificial intelligence can feel like they come from two different planets, one heavy and physical, the other abstract and new. Getting to stand in the middle and make that connection feel natural, even a bit exciting, is a rare thing. When someone walks up afterward and says “I could actually use this in my plant,” that's the whole point. That's the moment you're there for.

And that's really what these events are. Not endpoints, but invitations. The best conversations started at our booth, after the slides were done. People coming over with real questions, real curiosity about how AI is quietly reshaping the way we work every day at PSI. If I did my job right, I didn't just describe a product. I helped a few people picture a calmer, less overwhelming way of dealing with the mountains of data their operations throw at them.

Bokaro reminded me why I've stayed in this field, and with this company, as long as I have. It's the mix of real problems, real people, and technology used with a bit of honesty and purpose. Getting to stand up and say that out loud, for PSI, on the ISA Kolkata platform — that one is going to stay with me for a while.