60% more sales, 75+ hours saved: how Codex changed a fleet management startup!

Hey everyone, it's me! Today I want to share a fun story about Codex being used on the sales floor!

OpenAI News openai.com

What was announced?

OpenAI News featured Proaction, a company that builds software for fleet management — everything from cars and trucks to construction machinery. Co-founder and COO Colin Knudsen shared how he's been using Codex, along with GPT-Live-1 and GPT-6 Astra, to streamline everything from sales demos to day-to-day operations.

The story so far

Because every fleet operates differently, showing prospects how Proaction's platform fits their specific business was essential to selling it. But building personalized demos required engineering time the team couldn't spare, so the founders had to rely on conversations and slide decks instead. As Colin puts it, "As a non-technical person, I used to have to loop engineers in if I wanted a demo."

What changes

With Codex, Colin now builds the demos himself. He feeds it the Granola call recording, prospect email threads, and any spreadsheets the prospect has shared, and Codex customizes an HTML demo environment that mirrors Proaction's product using the customer's own fleet. When he shares his screen, prospects see their own vehicles and equipment, organized the way they actually work — and can point out what needs adjusting on the spot.

Colin estimates these custom demos have increased the share of deals moving from initial contact into solution development (rather than nurture) by 50-60%. Each demo takes 30-45 minutes to build, and he makes four to six a month. Since an engineer would need about 10 hours to build a comparable demo, that's roughly 40-60 hours of engineering time saved every month. The demos also double as a reference for engineers once a prospect converts, cutting down on back-and-forth about requirements.

Dive Deep

Proaction also built a customer solution center with Codex, where prospects can log in and explore workflows and materials tailored to their business. Colin, who juggles sales, customer support, and product management, uses Codex plugins for Granola, Gmail, Slack, Linear, GitHub, and HubSpot to pull together customer context and act on it — preparing follow-ups, creating Linear issues, updating HubSpot opportunities, and even running a scheduled automation that reviews recent calls and drafts sales updates. With 15-20 distinct tasks a day, he estimates Codex saves him 25-33 hours a month.

Proaction uses OpenAI models elsewhere in its platform too. When customers submit photos of vehicle issues, ChatGPT-5.6 Sol helps identify the damage. With GPT-Live-1, the company is building agents — its "Managed Execution Layer" — that handle more of the day-to-day work of running a fleet. One agent, Marty, coordinates vehicle maintenance: it talks with drivers about a problem, calls repair shops, arranges service, and helps get estimates approved and paid, with Proaction's team stepping in when human review is needed. On GPT-6 Astra, Head of Product Danny O'Halloran notes that "Astra's computer-use runs are more succinct" compared to GPT-5.6 Sol for the same work.

Wrap-up

• Codex lets non-engineers at Proaction build customer-specific demos themselves
• Custom demos increased deals advancing to solution development by 50-60%, saving 40-60 engineering hours a month
• Codex plugins centralize sales, support, and product work, saving Colin 25-33 hours a month
• Fleet-management agents like "Marty" are being built on GPT-Live-1 and GPT-6 Astra
• A great example for small teams looking to streamline sales and support without adding headcount!