NVIDIA saved 16 hours a week on internal work with ChatGPT Work!

Hey there, it's me! Today I want to share a case study from OpenAI News about how NVIDIA teams use ChatGPT Work internally. I'm curious how it actually plays out in practice, so let's dive in!

OpenAI News openai.com

What was announced?

OpenAI News featured two NVIDIA teams and how they've built ChatGPT Work into their work: Will Daney on the Go-To-Market (GTM) team, and Rachita Jain on the solutions architecture team.

The story so far

Will supports planning for GTC, NVIDIA's global AI conference, and previously did the account-list building and registration tracking by hand in spreadsheets. During the lead-up to the event, that manual analysis alone consumed about 40% of his time.

Rachita faced a similar challenge: keeping up with new models, benchmarks, and research appearing daily across the AI industry, and figuring out which developments actually mattered to NVIDIA's internal projects. The information itself wasn't hard to find — connecting it to the right context was.

What changes

Will turned his manual process into a ChatGPT Work workflow that now runs twice a week, saving about 16 hours per week across the 12-week GTC planning cycle.

"With ChatGPT, I think the real key is that I'm able to take a workflow I've already developed and I'm able to automate it event over event with little to no overhead."

—Will Daney, Go-To-Market Strategist

Rachita built a workflow that reviews trusted external sources alongside internal context, and now distills roughly 25-40 external AI updates into 5-8 actionable signals every week.

Dive Deep

For Rachita, ChatGPT Work isn't just for gathering information — she can go from an idea, through exploring approaches, working through a codebase, and debugging, all in the same environment. In one case, that took a prototype from idea to working product in about 3-5 days, versus an estimated 2-3 weeks if she'd stitched the work together across separate tools.

Removing the friction of switching between disconnected tools is what let a side project turn into something shipped in days rather than weeks.

Wrap-up

• NVIDIA's GTM team automated GTC event prep, saving about 16 hours a week
• The solutions architecture team now turns 25-40 weekly AI updates into 5-8 actionable signals
• Idea generation, coding, and debugging in one environment cut prototype time from 2-3 weeks down to 3-5 days in one case
• Teams can share their own workflows with others, who then customize them for their region

If you're looking for real examples of AI adoption inside a large org, or thinking about how to automate a recurring workflow on your own team, this one's worth a read!