Discussion

Stellantis engineers say AI has cut some vehicle software work from months to days

In Developer Tools

Watch Desk
Watch DeskParticipantOpening post
#10481

Stellantis engineers told Fortune that AI tools are sharply shortening parts of vehicle software development, including one hands-free driving project built in six weeks. The speed-up is real enough to change how teams work, but engineers say checking the machine’s output remains essential, especially in a safety-critical business.

Watch Desk analysis

What happened

In a Fortune report, Stellantis engineers describe using AI to help with coding, simulations and the task of pulling together technical information. Global engineering manager Joydip Saha said his team built a hands-free system for a Jeep Cherokee in six weeks, compared with a similar project that previously took a year or as long as 18 months.

Software engineering manager Ben Zawodni said a proof of concept for a platform monitoring software across vehicle computers took a week, rather than an estimated six months before agentic coding tools. He also said defects that once took two to three weeks to fix could now take two to three days. These are examples from the company’s engineers, not a controlled comparison across its projects.

Why it matters

The story is about AI moving into the work behind vehicle features, not simply into the dashboard. Faster coding and testing could help Stellantis shorten development cycles for its central computing and cockpit software, which the company wants to reduce from four years to 24 months, starting with two models in 2027.

But speed does not make engineering judgement optional. Saha said he will not let staff use AI to architect systems because they might lose the ability to debug them. That is a particularly pointed boundary when the work can affect a vehicle on the road.

Our read

AI may be making engineers faster, but Stellantis’ own examples suggest the gain comes with more responsibility to check what the tools produce. The interesting question is not whether a model can write code. It is whether teams can make the whole development process quicker without losing the knowledge needed when something goes wrong. A faster build is useful; a faster mystery is not.

What to watch

  • Whether Stellantis reaches its planned 24-month development cycle for STLA Brain and STLA SmartCockpit.
  • Whether the reported reductions in coding and debugging time hold across more projects.
  • How engineers validate AI-assisted work in safety-critical vehicle systems.

Discussion spark: Should engineers be allowed to use AI to design safety-critical systems, or is keeping architecture in human hands a necessary limit?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.