PM Stack Daily logo

PM Stack Daily

Archives
Log in
Subscribe
September 25, 2026

OpenAI's second executive exit this week, and a CRO hire to fix sales

Issue #015 · 4 min read

OpenAI's second executive exit this week, and a CRO hire to fix sales

Plus Gemini 3.7 Flash ships three weeks after 3.6, and GPT-5.6 gets 14x faster

The big story

OpenAI just lost its second executive this week.

Denise Dresser is out as chief revenue officer after nine months on the job, and OpenAI has already named her replacement: Dali Rajic, who was president and COO at Wiz — that's TechCrunch.

Nine months is not a tenure. It's barely enough time to learn the org chart.

The company's own announcement frames Rajic's job as helping "businesses realize the full value of AI" — OpenAI's post is worth reading for what it doesn't say about why Dresser left.

The Verge frames this as the second departure in a single week.

That's the number that matters here, not the title on the door.

If you sell into enterprises and OpenAI is one of your vendors or competitors, a revolving door at the top of sales is worth watching. Deals in flight, pricing conversations, and champion relationships all get disrupted when the person who owns the relationship changes twice a year.

What shipped

Gemini 3.7 Flash landed three weeks after Gemini 3.6 Flash, with Google claiming "substantial improvements" — Ars Technica has the release cadence.

It's already rolling out in GitHub Copilot, where GitHub says early testing shows gains in web and app development and agentic tasks.

Vercel has it too, at 50% off through AI Gateway.

Three weeks between major versions is a pace that makes evaluation cycles obsolete before they finish. If your team benchmarks models before adopting them, you're now benchmarking a moving target.

OpenAI previewed Ultrafast, a mode that runs GPT-5.6 Sol up to 14 times faster, hitting 750 output tokens per second by running on Cerebras hardware instead of OpenAI's usual stack — see OpenAI's announcement and TechCrunch's writeup.

This is aimed squarely at enterprise buyers who need latency guarantees, not chatbot users. If you have a product where response time is the thing users complain about, this is worth a test call before you assume it's marketing.

Linear added team-led initiatives, letting a specific team own a strategic initiative rather than leaving ownership implicit — the changelog also adds private-team initiatives for work that needs tighter access.

Small feature, real problem: "whose initiative is this" is a question every cross-functional roadmap eventually has to answer, and most tools make you answer it with a doc nobody updates.

What I'd actually do this week

Pull up your vendor list and check how many OpenAI relationships run through a single named contact. If Dresser's exit affected a deal you're in, get ahead of the handoff instead of waiting for your rep to explain it.

Run one real task through GPT-5.6 Sol's Ultrafast preview if latency has ever come up in a customer complaint about your AI feature. Fourteen times faster is either genuinely useful or it's a benchmark number that doesn't survive contact with your actual workload — worth twenty minutes to find out which.

If you're on Linear, try assigning a private team to lead one initiative this week and see whether it actually clarifies ownership or just adds a field nobody fills in.

Reply and tell me what you're seeing — especially if you've already run anything through Ultrafast.


Tools mentioned

  • Gemini 3.7 Flash in GitHub Copilot
  • Gemini 3.7 Flash on Vercel AI Gateway
  • OpenAI Ultrafast preview
  • Linear team initiatives

Read this issue on the web  ·  PM Stack Daily

Don't miss what's next. Subscribe to PM Stack Daily:
← Newer OpenAI disbanded the team meant to catch this exact problem Older → Twitch will train AI on your stream unless you say no
Powered by Buttondown, the easiest way to start and grow your newsletter.