Fleet-management software company Proaction has restructured how it sells and supports customers using OpenAI's Codex, GPT‑Live‑1 and GPT‑6 Astra, according to a case study published by OpenAI.
Proaction's platform helps businesses manage vehicle and equipment fleets, and since every fleet operates differently, personalized demos were central to closing deals — but building them previously required scarce engineering time. Co-founder and COO Colin Knudsen now builds four to six customized, interactive demos a month directly in Codex, each taking 30 to 45 minutes, using context pulled from Granola call recordings, prospect emails, and shared spreadsheets. He estimates this avoids 40 to 60 hours of engineering work monthly, since comparable engineer-built demos took about 10 hours each.
Knudsen says the share of deals advancing from initial contact into solution development, rather than stalling in nurture, has increased 50 to 60 percent since introducing the custom demos. When prospects convert to customers, engineers receive the demo as a visual reference, reducing clarification back-and-forth.
Beyond sales, Knudsen uses Codex plugins for Granola, Gmail, Slack, Linear, GitHub and HubSpot to consolidate call follow-ups, issue creation and opportunity updates in one interface, plus a scheduled automation that reviews recent calls and drafts sales updates. He estimates this saves him 25 to 33 hours a month.
On the product side, Proaction uses GPT‑Live‑1 to power what it calls a Managed Execution Layer — voice agents that act on fleet tasks rather than just tracking them. One agent, Marty, talks with drivers about vehicle problems, calls repair shops, arranges service, and helps get estimates approved and paid, with Proaction staff intervening when human review is needed. GPT‑6 Astra is also used for faster computer-use tasks; Head of Product Danny O'Halloran said Astra's runs are more succinct than prior models for the same work.