Proaction builds software for businesses that operate fleets of vehicles, from cars and trucks to construction machinery. Because each fleet works differently, personalized demos are central to sales, but they used to require engineering time the founders could not spare. Colin Knudsen, co-founder and COO, says Codex changed that: as a non-technical person, he used to loop engineers in for a demo; now he builds them himself. He creates four to six customized, interactive demos a month, each taking 30 to 45 minutes. After a sales call, he points Codex at the Granola recording, email threads, and spreadsheets a prospect shared, and Codex generates an HTML demo environment mirroring Proaction’s product and that customer’s fleet. Prospects see their own vehicles and workflows, can point out what to adjust, and help shape the solution without engineering involvement. Knudsen estimates the percentage of deals moving from initial contact into solution development, rather than nurture, has increased by 50% to 60%.

The demos also free engineering time. Knudsen estimates a comparable hand-built demo would take about 10 hours; at four to six demos per month, that is 40 to 60 hours of engineering work avoided each month. When a prospect becomes a customer, the customized demo becomes a visual reference that reduces back-and-forth about what to build. Proaction also used Codex to build a customer solution center where prospects can log in, explore workflows tailored to their business, and review sales materials. That gives customers a way to explain what they need and gives non-engineering teammates a path from those conversations to clearer requirements before engineers get involved.

Codex is also Knudsen’s daily work hub across sales, support, and product management. Through Codex plugins for Granola, Gmail, Slack, Linear, GitHub, and HubSpot, he pulls call transcripts and email history to prepare follow-ups, creates Linear issues, and updates HubSpot opportunities. A scheduled automation reviews recent calls and prepares sales updates for the team. He estimates he has 15 to 20 distinct tasks a day and that Codex saves him 25 to 33 hours a month, replacing the previous routine of jumping between tabs and copying information. Combined with the engineering hours, reported savings are roughly 65 to 93 hours per month.

Beyond Codex, Proaction uses OpenAI models across its platform. ChatGPT-5.6 Sol helps identify damage when customers submit photos with vehicle issue reports. With GPT-Live-1, Proaction is building what it calls a Managed Execution Layer so agents can execute work for customers rather than only help track it. Agents built on GPT-Live-1 and GPT-6 Astra make voice calls, review documents and images, analyze text, and respond in chat. One agent, Marty, coordinates vehicle maintenance: it talks with a driver about a problem, calls repair shops, arranges service, and helps get estimates approved and paid, with Proaction staff stepping in when human review is needed. Head of Product Danny O’Halloran says GPT-6 Astra’s computer-use runs are more succinct, while the same work with GPT-5.6 Sol required a much longer run. The reported net effect is more time to sell, support customers, and move the business forward, and more time back to fleet managers through voice-enabled agents.