The exact 1,400-token system prompt, branch logic, and SMS receipts that took one Phoenix HVAC shop from 38% after-hours capture to 91%.
Most HVAC voice agent demos sound great on stage and fall apart in February at 11pm when a frantic mom calls because her furnace died and her toddler is sleeping in a 51-degree room. The agent loops, mispronounces her zip code, quotes a dispatch fee that does not exist. She hangs up and calls the next shop on the Google list. This is the script we wrote after listening to 312 real recordings from a Phoenix shop to fix that. Why this is harder than it looks The most common failure mode with voice for hvac businesses is treating the problem as a model selection problem.
The hard parts are the data pipeline feeding it, the eval that catches regressions, and the human ownership layer that keeps the system honest after the implementer leaves the building. We have shipped this category of system enough times to recognize a few patterns. The teams that win allocate roughly 20 percent of project time to the model and prompts, 40 percent to data and integrations, 25 percent to evals and observability, and 15 percent to change management. The teams that lose flip those numbers, spend 70 percent on prompts, and end up with a great demo that nobody trusts. The good news is that none of this is novel engineering.
The exact 1,400-token system prompt, branch logic, and SMS receipts that took one Phoenix HVAC shop from 38% after-hours capture to 91%. It is filed under AI Voice because that is where operators looking for this problem actually start, and it is written from production work rather than from a content calendar.