AutoScout24 cut some projects from 2-3 weeks to 2-3 days with Codex!

Hey there, it's me, Shii-chan! Today I dug up a real-world story about how a big company rolled AI out across its entire engineering org. Let's go!

OpenAI News openai.com

What was announced?

Over on OpenAI's News, there's a customer story about AutoScout24 Group, one of the largest car marketplaces in Europe and Canada, scaling its engineering with AI-powered workflows.

AutoScout24 connects more than 30 million monthly users with over two million vehicle listings. It runs brands like AutoScout24 in Europe and AutoTrader.ca in Canada, supports a network of 45,000 dealer partners, and employs around 2,000 people worldwide.

The gist: the company paired ChatGPT and Codex to rethink how software gets built.

The story so far

As product expectations rose and systems grew more complex, AutoScout24 was under pressure to ship faster without hurting reliability. Between large-scale migrations, legacy systems, and rising engineering demand, incremental improvements were no longer enough.

The arrival of large language models gave the team a timely chance to rethink how software is built, tested, and scaled from the ground up.

What changes

The headline is speed. According to the article, development timelines dropped from 2-3 weeks to 2-3 days for select projects. That's a huge gap!

Engineers spend less time on manual work like reviews and documentation, so they can iterate and experiment faster. Even non-engineers can now prototype and validate ideas on their own, which widens the innovation capacity across the org, and that ultimately means better platform improvements for buyers and dealers alike.

Here's how CTO Frederik Kraus put it:

"Codex has emerged as a key enabler in our engineering workflows, delivering measurable impact in productivity, quality, and speed."

Dive Deep

AutoScout24 went with a dual-layer rollout.

First, ChatGPT went company-wide, giving about 2,000 employees access to AI tools and building a strong baseline of AI literacy.

In parallel, Codex was embedded into engineering, data, and product workflows, putting a coding agent directly in the hands of around 1,000 builder employees. Codex was chosen after a three-month evaluation across teams, where it showed strong results in usability, workflow fit, and measurable gains in productivity and code quality.

To make adoption stick, the company set up a cross-functional "AI Champions" network. It created a feedback loop between central leadership and individual teams, translating AI capabilities into practical use cases and embedding AI into existing workflows instead of bolting it on as a standalone tool.

Codex proved useful across several high-impact use cases:

• Automated pull request reviews
• Large-scale refactoring
• Technical documentation
• Post-incident analysis

Beyond engineering, AI tools let non-technical roles prototype and validate ideas independently, speeding up innovation across the whole organization.

The leadership lessons the article shared: combine broad AI access with deep workflow integration, prioritize real-world use cases over top-down mandates, use cross-functional champions to scale knowledge organically, evaluate AI tools with measurable engineering metrics, and focus on augmenting, not replacing, your existing teams.

Wrap-up

• OpenAI's News shared AutoScout24 Group's AI adoption story
• A dual-layer rollout: ChatGPT for ~2,000 employees, Codex for ~1,000 builders
• Codex was adopted after a three-month evaluation, shining at PR reviews, large refactors, docs, and post-incident analysis
• Select projects went from 2-3 weeks to 2-3 days in development time
• An "AI Champions" network was the key to making adoption stick on the ground

If you're a tech leader wondering how to embed AI across a large org, or an engineer thinking about improving your dev flow, this story is packed with hints!