Thirteen years on one question: what should this cost?

I started out doing pension calculations at Aon — the kind of work where a spreadsheet quietly becomes load-bearing and nobody notices until it breaks. I automated a lot of it in VBA, and somewhere in there I figured out that I was more interested in the tool than the calculation. That turned out to be the whole career.

Since 2012 I’ve built enterprise pricing and revenue management software, most of it at PROS. I came in on revenue management products, spent a stretch owning platform work — user management and single sign-on across every application on the platform, which reached 90% adoption — and then moved into transportation and logistics, where I owned guided-selling workflows and the APIs that connected live logistics data to pricing and optimization models.

In 2024 I went somewhere different on purpose. At Allied Outdoor Solutions I owned the entire technology roadmap for a company whose business isn’t software, which meant making the case for investment to executives who were rightly skeptical, and sequencing work across customer experience and operations at the same time. It was the first time I had to argue for the whole portfolio rather than one product in it. I’d recommend the experience to any PM who has only ever worked at software companies.

Now I’m at XPO, leading product for AI-driven pricing optimization on LTL freight. I work with data science teams on models for dynamic pricing, rate recommendation, demand forecasting and margin optimization, and with engineering on getting those models into production systems that have to answer in real time.

What I actually believe about this work

A pricing model that nobody trusts doesn’t get used, and a model that doesn’t get used has no value no matter how good the offline metrics are. Most of the hard work in AI-driven pricing isn’t the model — it’s the surface around it: what the seller sees, how much room they have to disagree, whether the recommendation comes with a reason. I care a lot about that surface. A studied computer science and mathematics background helps me have a real conversation with the data scientists; thirteen years of watching enterprise software get adopted or ignored is what tells me where to point them.

Outside of that

I took two years off between 2016 and 2018 to travel internationally, and came back to a broader role than the one I’d left — the best professional decision I’ve made, and the reason I speak Spanish well enough to be useful in it. I’m based in Houston. I keep a small collection of self-hosted services running, mostly because operating software you depend on teaches you things that reading about it doesn’t.

The fastest way to reach me is tompatella@gmail.com.