Artificial Intelligence brilliantly handles the most burdensome of human jobs to be done, yet it cannot escape the old adage that you are what you eat.
As we sit on the precipice of launching a powerful new agentic suite, Stella, I shall take a nanosecond to reflect on our journey. We have steadily launched six Smart Assistants over the course of the past year, with over 80% user adoption. The launch of Stella AI Agents feels different. It literally eats 14 years and nearly 50 million records of commercial contracting data across thousands of customers for lunch and counting! Trade Intelligence is how we take the deep operational history we already have—what’s happened across assets, sites, technicians, inspections, deficiencies, quotes, outcomes—and convert it into usable intelligence inside the workflow. Not as generic “AI,” but as domain-specific understanding that’s tied to commercial service realities.
Our AI journey has accelerated this year, and so has our understanding of how to effectively use AI in ways that drive real financial outcomes. There is a lot of market pressure to launch AI solutions. But time-to-market doesn’t always equate to value-to-market. We are keenly aware that both are important.
There is equally as much consternation as to whether or not we are in an AI bubble. We may or may not be in a bubble, but that is irrelevant to the fact that regardless of the market value attributed to artificial advancements, there is unequivocally enormous operational efficiency and revenue acceleration possible, in the commercial trades especially — an industry that is operationally heavy and physically resilient.
Stella automates key phases of the service lifecycle — contracting, scheduling, servicing, quoting, invoicing, and collections – for HVAC, fire protection, and mechanical contractors. Phases that are most correlative to revenue production.
What Does AI for Commercial Service Actually Mean?
AI for commercial service means technology that is embedded directly in the workflows commercial contractors use every day — not a standalone tool you have to leave your system to access. It means AI trained on domain-specific data: job history, asset records, deficiency reports, and service contracts — not generic internet content. And it means AI that can be measured against real business outcomes: quotes going out faster, approval rates improving, margins protected, and technician time reclaimed.
What commercial service actually demands from AI
In the trades, execution is the business. A field service organization lives and dies by the quality of its workflows — how quotes get built, how techs get dispatched, how deficiencies get documented and followed up on. These are not abstract problems. They are the mechanics of how revenue is created or lost every single day.
That operational reality sets a higher bar for AI. A general-purpose tool that generates text or summarizes notes might be useful somewhere. But in commercial service, useful means embedded — in the workflow, in the platform, and grounded in the kind of industry-specific context that reflects how work actually moves from one step to the next.
When AI is designed to that standard, it is not a novelty. It becomes part of how work gets done.
Starting with the right questions
Before we wrote a single line of AI code, we asked: What are the most valuable jobs-to-be-done in a commercial contracting operation? Which activities in the service lifecycle most directly impact revenue and cash? And for which of those activities could an AI model — trained on real, trade-specific intelligence — perform the work well enough that our customers’ increasingly scarce and valuable human resources could focus on more strategic work?
Those questions shaped everything. They kept us grounded in the operational reality of our customers.
Consider this example: a commercial contractor with 400 open deficiencies. Each one has to be located, reviewed, priced, and turned into a quote before it can become revenue. In many operations, that process takes 20–30 minutes per deficiency — involving printed reports, multiple tabs, and manual data entry. Multiply that across a backlog of hundreds, and you have a revenue bottleneck that has nothing to do with a lack of identified work. AI for commercial service, done right, compresses that cycle dramatically — without sacrificing accuracy or requiring your team to learn a new system. Reducing average quote turnaround by two weeks increases quote approval by 13.3%, and generates millions of dollars in new revenue.
This is the real value we are delivering.
We are also just as focused on the efficacy and validity of our models. We are thoughtful about defining constraints — the actions are models are not allowed to perform. We take care in articulating success criteria and delivering observability. We also realize that it is important to offer user experiences that build trust in our models and their results.
We have also learned along the way. We have learned that we can build agentic capabilities faster when we have an artificial intelligence platform with patterns. This includes determined ways of accessing and writing data, authorizing users, persisting data, and automating testing.
I am so excited to show the world how Stella is a thoughtful, and trusted way of leveraging AI to advance the billion dollar, resource constrained commercial contracting industry. Instead of making promises we can’t keep, or trying to enter new markets we can’t serve — we have kept our heads focused on creating value for our customers.
