
Our Mladen Subasic, together with Ted Stoie and Joe Fiella of Bridge Specialty Group, shared what three years of submission automation and disciplined AI looks like at scale.

In the third episode of Bound & Beyond, two senior leaders from Bridge Specialty Group, Ted Stoie and Joe Fiella, joined Mladen Subasic, Chief Product Officer at OIP Insurtech and BoundAI. The conversation covered a broad range of topics, including submission automation, change management, build vs buy, and where AI actually does and doesn't deliver. AJ Beatovic hosted the podcast.
Bridge Specialty Group has grown almost entirely through acquisition, and that history shapes how it thinks about standardization. The tension between efficiency and local autonomy is largely structural.
“We hold that tension very much," Ted said. “More than anything, it's about listening to the local offices and hearing where they're struggling, where their gaps are, and then spotting the common threads that we can then go after.”
Joe added that the pace of AI and automation has made this tension much harder to ignore. "You can't go build it 30-something or 40-something different ways by office. You've got to scale it. But you can't lose that entrepreneurial selling gene that our offices have."
Three to four years ago, Bridge decided to implement a submission automation engine in conjunction with OIP and BoundAI. That was an early move for them and definitely a conscious decision.
Joe was very clear on the original driver. "The one pain point we had was getting that data. What you don't want to do is go to the front line and ask our people who are sitting at the desk to go and enter all that data. They're already very busy."
Ted (who joined post-launch) was very much aware of the downstream impact. "You just can't imagine the position we'd be in as a leadership team if we hadn't had that insight. All the other things we're talking about in terms of what we'd like to do and build out the roadmap from a technology perspective at Bridge, I can't think of one that doesn't in some way be connected to the submission engine."
Mladen pointed out what made the initiative work from the product side. "Your problem was a workflow problem. You didn't want to go out and try something new and see if it worked. You actually had a genuine thing that you wanted to solve, and you wanted to solve it before the market."
Adoption across a decentralized organization with dozens of offices wasn't immediate. The discussion on that point was very frank.
"There was a very wide distribution of uptake initially," Ted explained. "Some offices were absolutely on board; others we saw virtually no uptake at all. We just couldn't have had those conversations without the ability to measure that."
Joe explained what it takes actually to get adoption moving. "You've got to win hearts and minds. You're ultimately thinking about people who have been through that process for 10, 15 years in many cases, and they're very good at their process. You've got to tell the story, you've got to tell people why we're using the technology, how to use it, and get them on board."
Mladen put it as a product-launch mindset. "If we treat the rollout of the technologies we have in our organization as a product launch to the market and give it the same care as you would to a product launch, then we're going to get all of those benefits. Adoption follows value that people can see and feel."
One of the more pointed exchanges in the conversation was about where organizations are spending their AI budgets and where they're actually seeing returns.
Mladen was very clear on the gap. "80% of the attention and budget is being applied to the AI side of growing the business. What we're actually seeing in terms of output is the reverse: 80% of the gains are on the operational execution side, 20% on the growth side."
Joe agreed. "It's easier to get the bang for your buck on the operational side as it's a workflow. You can plug AI into various parts of that workflow to make it quicker and faster, and you can measure that. On the growth side, it's very difficult to quantify."
Build versus buy (and why the question is changing)
The discussion ended on the build v buy issue with both Ted and Mladen clearly unhappy with the way it had been framed.
“Build versus buy is utterly myopic when you're talking about a technology solution," said Ted. "Technology is absolutely pointless if it's not adopted. What should be the key thing is how quickly we can get the benefits and how easy it is to sustain those benefits over time.”
Mladen went on. "It's not build or buy anymore at all. Your alpha is whatever makes people come to you rather than the shop down the road. That's where you need to be investing. Everything else you hire."