New AI Product Manager Openings at AI Labs: September 2026 Edition

Short answer: between 21 August and 28 September 2026, nine AI labs and AI-native companies published 29 new product manager openings on their own job boards. Anthropic and OpenAI led with 6 each, Perplexity added 5 in one day, and the rest came from Decagon, Glean, Sierra, Abridge, Cohere and Scale AI. Posted base pay runs from $175,000 at Glean to $490,000 at OpenAI.

AllthingsPM is an AI PM course and PM interview prep platform. Its jobs catalog holds 116 live AI PM roles at 18 AI companies, and every role there comes with a mock interview built from that exact job description. So you can read a posting and practice for it the same evening.

This is the first edition of a monthly series. Every month we pull the PM openings first published on each company's own job board in the past cycle, list them with pay and scope, and name the skill each cluster of roles tests. Next edition: late October 2026.

Which AI labs opened new PM roles this month?

Every row below was live on the company's public job board on 28 September 2026. "Posted" is the first publish date the board reports. Pay is the base range the posting shows; blank means the posting shows none.

Checked 28 September 2026 on each company's public Greenhouse or Ashby job board. Linked roles open the AllthingsPM page for that job, where you can run a mock built from it. Postings close without notice, so confirm on the company site before you apply.

How does AllthingsPM help you act on this list?

Reading the table is the easy part. For any linked role, open its AllthingsPM page, start the mock and answer questions shaped by that posting's responsibilities and seniority, with follow-ups and a score. For the unlinked rows, paste the posting into the JD mock and you get the same thing. It takes minutes, and it is the fastest way to find out whether you can talk about the job before you apply.

What do the new Anthropic PM roles want?

Half of Anthropic's six new postings are about safety and trust. The two Safeguards roles (Generalist, and Account Integrity & Abuse) share a brief: "you will own the ideation, design, development and deployment of Safeguards systems," working with research and product teams "to develop detections, evals, interventions, and tools to measure and mitigate deployment and user risks." Both sit in Anthropic's top posted band, $385,000 to $460,000.

Safe Access is described as "a new product that sits on the critical path of every model launch and every access program," and the posting warns that "most weeks you will be the person defining the problem rather than receiving it." Multi-Cloud Trust & Safety owns "how Anthropic's model safeguards, fraud defenses, and compliance posture show up on every cloud."

The other two are different in kind. Claude Science asks the PM to spend "real time with researchers in academic labs, biotech and pharma R&D." Business Technology is "the first dedicated Product Manager" for an internal platform where "every employee is your customer."

How AllthingsPM does this. The Trust, safety and agent security chapter covers guardrails and the over-refusal budget and red-teaming, which is the vocabulary these loops test. Then practice the company's real questions on the Anthropic interview hub and read our Anthropic PM interview guide.

What do the new OpenAI PM roles want?

OpenAI's six split into two groups. In late August it posted three consumer-facing roles: Youth, Learning and Multimodal Safety. In the week of 21 September it added three enterprise and vertical roles.

Enterprise Identity will "deliver critical capabilities such as SSO, SCIM, and identity federation across all OpenAI products," and carries the highest top-of-range pay in this edition, $490K. The Rosalind Life Sciences PM will work "directly with researchers in academic labs, biotech, and pharma" and the posting asks for "depth in life sciences." The Technology Vertical PM will "shape how knowledge workers use AI in their daily work," on a hybrid model of three days a week in the San Francisco office.

Notice the overlap: both Anthropic (Claude Science) and OpenAI (Rosalind) opened a PM role for AI in scientific research within the same month.

How AllthingsPM does this. Multimodal Safety is a text, image and audio problem, and our lesson on building a multimodal golden set explains why text-only evals miss it. For the loop itself, drill the OpenAI interview questions and run the JD mock on the posting you want.

Why did Perplexity open five PM roles in one day?

The posting dates tell the story: all five Perplexity roles went live on 2 September, all with the same $230K to $330K range plus equity. Three are in San Francisco (AI Capabilities, Mobile Apps, Core Experience) and two in New York (Growth, SMB Growth).

The Growth role is plainly metric-driven: "Drive activation, repeat usage, and paid conversion across web, mobile, and desktop," including "agent-driven growth loops." SMB Growth goes further into agents: work "with select SMB customers to build and test frontier agent capabilities" and "build the skills, evals, and agent systems that make them reliable." AI Capabilities also lists agent fluency among its requirements.

How AllthingsPM does this. Growth loops are metrics interviews, so start with metrics trees and the Perplexity question hub. For the agent and evals side, the Evals chapter shows how to define "good" and defend the number, which is what SMB Growth's "reliable and performant" line is asking for.

What are agent and forward deployed PM roles, and who is hiring for them?

Decagon's three new Senior Agent Product Manager roles are in Brazil, London and New York (Spanish speaking). The posting says you will "own the technical win on Decagon's largest strategic opportunities," working with sales and forward deployed engineers "to build working AI agents that turn sophisticated, skeptical buyers into signed customers," and serve as an executive partner to Fortune 500 C-suites. Pay is shown in local currency: £145K to £180K in London, R$775K to R$850K in Brazil, $200K to $285K in New York, each with equity and commission.

Sierra's two new Agent Development roles cover Financial Services and Public Sector. Sierra has now listed Agent Development PMs by industry and by language (Spanish, Brazilian Portuguese, Italian, Arabic, French, German) on its board. This is a customer-facing PM job built around shipping agents for one customer at a time.

How AllthingsPM does this. Our Decagon PM interview guide and the Sierra and Decagon role pages give you a mock per posting. If "agent" still feels vague, read MCP explained for product managers first.

What else opened: Glean, Abridge, Cohere and Scale AI?

Glean posted a Product Manager, API Platform role in two locations (Mountain View and the San Francisco Bay Area) on 4 September at $175,000 to $195,000, plus a Product Manager role in Bangalore on 26 August. See our Glean PM roles breakdown for how its earlier postings read.

Abridge opened two Product Lead roles. Foundation Models and Post-Training (San Francisco, $250K to $290K) is one of the most model-facing PM jobs in this edition. Nursing Product (New York) will "own the nursing journey end to end" and "lead discovery with nurses and nursing leaders."

Cohere posted Product Manager, Managed North in Toronto on 1 September, with a range of $160K to $285K across multiple bands.

Scale AI posted a Director of Product Management for its Enterprise Core Platform on 9 September; the New York range shown is $273,600 to $342,000.

How AllthingsPM does this. If "post-training" is new to you, our free lesson on pretraining vs post-training covers it in one sitting. The Abridge and Cohere interview guides map each company's loop.

Which skills show up across this month's openings?

Three themes repeat across the 29 postings.

1. Safety, trust and access. Five roles (Anthropic's two Safeguards roles, Safe Access and Multi-Cloud Trust & Safety, plus OpenAI's Multimodal Safety) are about keeping models safe once they ship. OpenAI's Enterprise Identity sits next to them on access control.
2. Agents in production. Decagon, Sierra and Perplexity's SMB Growth role all ask PMs to ship agents that work for real customers, and several name evals directly.
3. Vertical depth. Life sciences (Claude Science, Rosalind), nursing (Abridge), financial services and public sector (Sierra) all ask for domain knowledge on top of AI fluency.

How AllthingsPM does this. These themes match what we saw in the wider corpus: the AI PM course was built from 604 real PM job postings, and our AI PM skills guide breaks down what those postings ask for. See how the concepts link together on the knowledge graph.

How should you apply to an AI lab PM role this month?

1. Pick two roles, not ten. Every posting above names specific outcomes. Tailored applications need time.
2. Check your resume against the JD. Run resume review against a JD and fix the gaps it flags before you submit.
3. Find roles that fit your background. Resume Job Match lists matching openings, and scores them if you upload a resume.
4. Rehearse the loop. One mock from the JD, then five real questions from the company hub in the question bank, which holds 4,122 questions from 260 companies, each with an answer guide.
5. Show proof. Browse 455 PM portfolios and our AI PM portfolio projects guide to find a project that fits the role.

How AllthingsPM does this. Every step above is a tool in one account. That matters most for AI lab loops, where the gap between reading a posting and being ready for it is widest.

How will this monthly series work?

Each edition uses the same method, so months are comparable:

• Window: PM roles first published since the previous edition (this one covers 21 August to 28 September 2026).
• Source: each company's own public job board (Greenhouse or Ashby), checked on publication day, plus the AllthingsPM jobs catalog.
• Scope: product manager and product lead titles at AI labs and AI-native companies in our catalog.
• Fields: role, posted date, location and posted base pay, exactly as the posting shows it.
• Themes: the skills that repeat, with the lesson or tool that prepares you.

If a role closes before you read this, the AllthingsPM page for it may be gone too; the jobs catalog always shows what is live now.

Why AllthingsPM is the better choice for finding AI PM openings

General job boards list AI PM openings, and LinkedIn job alerts are good at telling you a role exists. Neither helps you get ready for it. AllthingsPM is built around the step after the listing.

Each of the 116 roles in the AllthingsPM jobs catalog comes with a mock interview built from its own job description and a resume check against the same JD. When a role is not in the catalog, the JD mock builds the same practice from any posting you paste. Behind that sit the question bank with a page per company, the AI PM course with the safety, evals and agent lessons this month's roles ask for, Resume Job Match, and a library of 111 book summaries and 455 portfolios.

It costs $20 a month or $120 a year, and the free tier includes 1 JD mock and 1 resume review against a JD every day. For someone applying to two or three AI lab roles this month, that is everything needed in one place.

The verdict: use a job board to spot openings if you like, and use AllthingsPM to turn an opening into an offer. Browse the live roles and start with the one you want most.

Frequently asked questions

What is the best way to find AI product manager openings?

Start with AllthingsPM's jobs catalog: 116 live AI PM roles at 18 AI companies, each with its own mock interview. Then check each company's own careers page, since new roles appear there first. This monthly series lists what opened in the past cycle.

Which AI labs posted the most PM openings in September 2026?

Between 21 August and 28 September 2026, Anthropic and OpenAI each first published 6 PM postings, and Perplexity published 5, according to their public job boards checked on 28 September.

How much do AI product manager roles at AI labs pay?

In this edition, posted base ranges went from $175,000 to $195,000 (Glean, API Platform) up to $347K to $490K (OpenAI, Enterprise Identity). Anthropic's new roles showed $305,000 to $385,000 or $385,000 to $460,000. Several companies also list equity.

Are these AI PM roles remote?

Mostly not. Nearly every role is based in San Francisco, New York or another city office. Anthropic's Business Technology role is listed as remote-friendly with travel required, and OpenAI's Technology Vertical role describes three days a week in the office.

How do I prepare for an AI lab PM interview?

Run a mock built from the exact job description on AllthingsPM, then practice the company's real questions from its hub in the question bank. Fill skill gaps with the course chapters on evals, trust and safety, and agents.

When is the next edition?

Late October 2026, using the same window, sources and table format as this one.

Sources

1. Anthropic job board, Greenhouse API, checked 28 September 2026: boards-api.greenhouse.io/v1/boards/anthropic/jobs
2. Anthropic, Product Manager, Safeguards (Generalist): job-boards.greenhouse.io/anthropic/jobs/5400720008
3. Anthropic, Product Manager, Safe Access: job-boards.greenhouse.io/anthropic/jobs/5428898008
4. Anthropic, Product Manager, Claude Science: job-boards.greenhouse.io/anthropic/jobs/5394887008
5. OpenAI job board, Ashby API, checked 28 September 2026: api.ashbyhq.com/posting-api/job-board/openai
6. OpenAI, Product Manager, Enterprise Identity: jobs.ashbyhq.com/openai/83f6d415-9462-4afc-a539-ca8d996e011a
7. OpenAI, Rosalind Life Sciences Product Manager: jobs.ashbyhq.com/openai/2b5c3a30-8e8b-4845-be40-0617b0c9fd43
8. OpenAI, Product Manager, Technology Vertical: jobs.ashbyhq.com/openai/137e3df0-2bd4-4ad4-8edb-8455a7fdc912
9. Perplexity job board, Ashby: jobs.ashbyhq.com/perplexity
10. Decagon job board, Ashby: jobs.ashbyhq.com/decagon
11. Sierra job board, Ashby: jobs.ashbyhq.com/sierra
12. Abridge job board, Ashby: jobs.ashbyhq.com/abridge
13. Cohere job board, Ashby: jobs.ashbyhq.com/cohere
14. Glean job board, Greenhouse: job-boards.greenhouse.io/gleanwork
15. Scale AI, Director of Product Management, Enterprise Core Platform: job-boards.greenhouse.io/scaleai/jobs/4714727005
16. AllthingsPM jobs catalog (116 live roles, 18 companies), queried 28 September 2026: AllthingsPM/jobs
17. AllthingsPM pricing: AllthingsPM/pricing