An AI product manager job description has seven parts: a mission line, the team and product, what you will do, what you bring, nice-to-haves, the pay range and the location. We built the template below by reading all 116 live PM job descriptions at 18 AI companies in the AllthingsPM jobs catalog, including Anthropic, OpenAI, Sierra, Scale AI and Glean. AllthingsPM is an AI PM course and PM interview prep platform, and every one of those 116 JDs comes with a mock interview built from its own text, so you can read a real posting and practice for it in the same place.
The short version: real AI PM JDs ask for customer work (87%), technical depth (78%), a roadmap (77%) and agents (69%) far more often than they ask for an MBA (17%) or SQL (3%). The median posting asks for about 6 years of experience, and half of them publish a salary range.
What does an AI product manager job description template look like?
Copy this. Each bracket is a line you replace. The notes under each section come from what the 116 real JDs actually do.
Job title: [Product Manager / Senior Product Manager], [Product area, for example Agent Platform, Evals, Enterprise]
About [Company]: [One or two sentences on the mission and what the company builds with AI.]
About the team: [What this team owns, who uses it, and why it matters now. Name the product surface: API, agent, enterprise app, model quality.]
What you will do
- Own the roadmap for [product area], from discovery to launch and iteration.
- Work directly with [enterprise customers / developers / end users] to understand their workflows and turn them into requirements.
- Partner with research and engineering to decide what the model should do, and define how you will know it works (evals, quality bars, success metrics).
- Ship [agents / AI features] into production and own their reliability, safety and cost.
- Write clear specs and communicate decisions across research, engineering, design, go-to-market and leadership.
What you bring
- [N]+ years of product management experience, ideally on [technical / developer / enterprise / AI] products.
- Strong technical fluency: you can read an API, reason about model behaviour, and go deep with engineers.
- Hands-on building: you prototype with AI tools yourself rather than relying only on process.
- Comfort with ambiguity and fast change.
- Excellent written and verbal communication.
Nice to have
- Experience with LLMs, agents, model evaluation or benchmarks.
- Experience selling to or deploying with [enterprise / regulated industry] customers.
- [Domain background: legal, healthcare, finance, security.]
Compensation: [Annual base salary range, for example $X to $Y], plus [equity and benefits].
Location: [City, office days per week, or remote policy].
Equal opportunity statement: [Your standard statement.]
Why this skeleton works
Generic PM templates stop at vision, roadmap and cross-functional work. Workable's popular product manager template, for example, lists a job brief, responsibilities and requirements, with a CS or engineering degree "preferred", and never mentions AI, LLMs or evals ([1]). The AI company JDs keep that frame and add three things: a named model or agent surface, a quality bar (evals), and proof that you build things yourself.
How AllthingsPM does this: every JD in our catalog of AI company PM roles is stored in full, with its title, level, location and posted pay range, so you can compare your draft against real ones in minutes. Open any role, such as Product Manager, Agent Development, Financial Services at Sierra, and a mock interview built from that exact text is one click away.
What do 116 real AI PM job descriptions actually ask for?
We ran a keyword pass over the full text of all 116 JDs, after removing the equal-opportunity boilerplate (it contains words like "evaluate" that would inflate the evals count). Here is how often each requirement appears.
| Requirement (phrase matched) | JDs | Share |
|---|---|---|
| AllthingsPM mock interview available | 116 | 100% |
| Customers | 101 | 87% |
| Technical depth (technical background, computer science, technical PM) | 90 | 78% |
| Roadmap | 89 | 77% |
| Enterprise | 85 | 73% |
| Agents or agentic | 80 | 69% |
| Ambiguity | 62 | 53% |
| LLMs, foundation or frontier models | 56 | 48% |
| Prototype, hands-on, builder | 48 | 41% |
| Metrics or KPIs | 39 | 34% |
| Degree or bachelor's | 39 | 34% |
| Evals, model evaluation, benchmarks | 27 | 23% |
| MBA | 20 | 17% |
| SQL | 4 | 3% |
Source: the 116 JDs in the AllthingsPM jobs catalog, read 26 September 2026. A keyword pass counts mentions, not hard requirements; a JD that says "MBA not required" still counts as mentioning an MBA.
Three things stand out. First, "customer" is nearly universal, because most of these companies sell to businesses: Sierra, Decagon, Glean and Harvey are enterprise products, and "enterprise" shows up in 73% of JDs. Second, agents (69%) are now more common than LLMs by name (48%). The product is the agent, and the model is a component. Third, the classic PM screens are fading. SQL appears in 4 JDs out of 116.
How AllthingsPM does this: the AI PM course is built around exactly these asks. There is a whole chapter on agents and agentic architecture, one on evals, and one on shipping into enterprise customers, because those are the words the postings keep using.
How many years of experience do AI PM job descriptions ask for?
103 of the 116 JDs state a minimum number of years. Taking the lowest number each one names, the median is 6 years.
| Minimum years asked | JDs | Share of the 103 |
|---|---|---|
| 3 or fewer | 8 | 8% |
| 4 to 5 | 38 | 37% |
| 6 to 7 | 37 | 36% |
| 8 or more | 20 | 19% |
The levels match. By title and text, 59 roles are plain PM, 34 are Senior, 10 are Staff, 8 are Lead or head of product, 1 is a Director and only 1 is an APM. AI companies are hiring experienced PMs and expecting them to learn the AI parts fast.
For your template, write the number you actually need. If you would hire a strong 4-year PM who ships agents, say "4+ years", not "6+". Anthropic's careers page makes the same point from the candidate side: "We care about what you can do, not where you learned to do it" ([2]).
How AllthingsPM does this: if your years are short but your skills are real, prove it. The JD resume review scores your resume against any JD line by line, and Resume Job Match finds roles your resume already fits.
Should an AI PM job description include a salary range?
In many places you have no choice. California requires employers with 15 or more employees to include the pay scale in every job posting, a rule in force since 1 January 2023 ([3]). New York State requires a pay range in postings from employers with four or more employees, effective 17 September 2023 ([4]).
In our catalog, 59 of 116 JDs (51%) list a pay range, from 9 of the 18 companies. Across those 59, the median bottom of the range is $210,000 and the median top is $290,000. The lowest bottom is $160,000 and the highest top is $460,000.
| Company | JDs in catalog | JDs with a pay range | Example posted range |
|---|---|---|---|
| Anthropic | 19 | 19 | $305,000 to $385,000 and $385,000 to $460,000 |
| Scale AI | 15 | 11 | $171,200 to $214,000 up to $210,000 to $345,000 |
| Glean | 14 | 11 | $160,000 to $240,000 up to $175,000 to $280,000 |
| Decagon | 10 | 7 | $200,000 to $285,000 and $232,000 to $290,000 |
| Harvey | 7 | 7 | $177,000 to $260,000 and $213,600 to $300,000 |
| Suno | 1 | 1 | $250,000 to $330,000 |
| Vercel | 1 | 1 | $172,000 to $258,000 |
| Midjourney | 1 | 1 | $185,000 |
Source: salary text on each JD in the AllthingsPM jobs catalog, 26 September 2026. OpenAI and Sierra roles in the catalog did not carry a pay range in the stored text.
Template rule: put a real annual base range in the posting body, not behind a link. It is the law in several states, and candidates filter on it.
How AllthingsPM does this: each role page in the jobs catalog shows the posted pay range next to the JD when the company publishes one, so candidates can compare levels and pay across AI companies before they apply.
Where are AI PM roles based?
Mostly in San Francisco. At least 66 of the 116 JDs (57%) name San Francisco, 11 name New York, 9 name London, and only 2 are listed as remote. If your role is remote, say so in the title line; it is rare enough to be a selling point.
How AllthingsPM does this: the catalog lists location on every role, and Resume Job Match searches wider job boards when you need roles outside these 18 companies.
What should the "What you will do" section say for an AI PM?
Name the product surface and the quality bar. The strongest JDs in the catalog do both in one line. An OpenAI life sciences PM role asks for someone who has "built model evaluations or benchmarks, ideally for scientific or agentic workflows" ([5]). A Scale AI public sector PM role asks you to "scope out model evaluation sets and performance requirements, consistently review results, and iterate on the solution" ([6]). A Glean AI quality role says the PM will "evaluate LLM models, define the roadmap for growing Glean's LLM portfolio" ([7]).
Good bullets for this section:
- Surface: "Own the roadmap for our customer-facing agent in [domain]."
- Customer: "Sit with enterprise customers during deployment and turn what breaks into requirements."
- Quality bar: "Define the evals and success metrics that decide when a model change ships."
- Safety and cost: "Own the reliability, safety and cost of the agent in production."
- Communication: "Write the spec and the launch memo that research, engineering and sales all work from."
Avoid bullets that could sit in any PM posting from 2015, like "manage the product lifecycle." The data says AI companies are screening for agents, evals and customers, so say that.
How AllthingsPM does this: the course lesson on the AI PM job now is free and walks through selection, taste and verification, the skills behind these bullets. The evals chapter teaches you to write the quality bar yourself.
What should the "What you bring" section say?
Keep it to five or six lines, and make each one testable in an interview. The catalog JDs lean on four:
- Years and domain. "5+ years of PM experience on technical or developer products."
- Technical fluency. 78% of JDs ask for technical depth in some form. Say what you mean: reading an API, reasoning about model behaviour, debugging with engineers.
- Hands-on building. 41% use words like prototype, hands-on or builder. Sierra puts it plainly: "you prototype, investigate, and work hands-on rather than relying solely on process" ([8]).
- Ambiguity and communication. 53% mention ambiguity.
Drop the MBA unless you truly require it (17% mention one). Drop SQL unless the role is analytics-heavy (3%).
How AllthingsPM does this: the course chapter PM as builder has you prototype and inspect an agent yourself, which is the "hands-on" line recruiters are now screening for.
How should candidates use an AI PM job description?
Read it as the interview script. Each "What you will do" bullet becomes a product sense or execution question, and each "What you bring" line becomes a behavioural one. A JD that asks for evals will ask how you would measure an agent. A JD that says enterprise will ask how you would roll out to a cautious customer.
A simple routine:
- Highlight every noun in the JD that names a product surface, customer or metric.
- Write one story from your past for each "What you bring" line.
- Run a mock built from that JD and see which lines you cannot answer.
- Fix the gaps with one lesson each, then mock again.
How AllthingsPM does this: step 3 is the JD mock. It reads the JD, asks questions in its language, follows up, and scores you. For company-specific questions, the Anthropic company hub collects real questions reported for that company, and our breakdown of Anthropic PM job descriptions reads their postings line by line.
Why AllthingsPM is the better choice for AI product manager job descriptions
Most JD template pages give you a generic skeleton and stop. Workable, Deel and Indeed publish solid general PM templates, and they are useful if you are hiring for a classic software role ([1], [9], [10]). None of them show you what AI companies actually wrote this year, and none of them help a candidate prepare for a specific posting.
AllthingsPM does both. The template above comes from 116 real JDs at 18 AI companies, not from memory. Each of those JDs lives on its own page in the jobs catalog with level, location and pay where posted, and each one has a mock interview built from its own text. If the role you want is not in the catalog, paste it into the JD mock and get a scored interview in text or voice.
Then close the gaps. The AI PM course was built from 604 real PM job postings, so its chapters on agents, evals and enterprise deployment map onto the lines these JDs repeat. The question bank has 4,122 real questions from 260 companies, each with an answer guide. The JD resume review checks your resume against the posting before you apply.
It is also the cheapest serious option: one free JD mock a day, and Pro at $20 a month or $120 a year. For hiring managers writing a JD and candidates reading one, AllthingsPM is the one place that shows you real AI PM postings and lets you practice against them.
Browse the 116 AI PM job descriptions and start a free mock.
Frequently asked questions
What is the best AI product manager job description template?
The best template is one built from real AI company postings. The AllthingsPM template above comes from 116 live PM JDs at 18 AI companies, and AllthingsPM lets you open each of those JDs and run a mock interview built from it. Generic templates from Workable or Indeed work for classic PM roles but do not mention AI, agents or evals.
What skills do AI product manager job descriptions ask for most?
In the 116 JDs we read, the most common asks are working with customers (87%), technical depth (78%), roadmap ownership (77%), enterprise experience (73%) and agents (69%). Evals appear in 23% and hands-on building or prototyping in 41%.
How many years of experience does an AI PM job need?
The median minimum in our 116 JDs is 6 years. 37% ask for 4 to 5 years and 36% for 6 to 7. Only 8 of the 103 JDs that state a number ask for 3 years or fewer.
Do AI PM jobs require an MBA or SQL?
Rarely. An MBA is mentioned in 17% of the 116 JDs and SQL in 3%. Technical fluency and hands-on building come up far more often.
How much do AI product managers earn, according to job postings?
Among the 59 JDs that post a pay range, the median range runs from $210,000 to $290,000 in base salary. Anthropic's posted ranges go from $305,000 up to $460,000. Equity and benefits are extra and vary by company.
How do I prepare for a specific AI PM job description?
Turn each bullet into a question, write a story for each requirement, then practice against the real text. AllthingsPM builds a scored mock interview from any JD you paste, and every role in its jobs catalog already has one.
Sources
- Workable, "Product Manager job description template": https://resources.workable.com/product-manager-job-description
- Anthropic, Careers page: https://www.anthropic.com/careers
- California Legislative Information, SB 1162 (pay scale in job postings, employers with 15 or more employees): https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202120220SB1162
- New York State Department of Labor, Pay Transparency Law: https://dol.ny.gov/pay-transparency-law-fare-grant ; effective date per Epstein Becker Green: https://www.workforcebulletin.com/new-york-states-salary-transparency-law-takes-effect-september-17-2023
- OpenAI, Rosalind Life Sciences Product Manager, via the AllthingsPM jobs catalog: /jobs/openai/rosalind-life-sciences-product-manager
- Scale AI, Product Manager of AI Applications, Global Public Sector, via the AllthingsPM jobs catalog: /jobs/scale-ai/product-manager-of-ai-applications-global-public-sector
- Glean, Product Manager, AI Quality, via the AllthingsPM jobs catalog: /jobs/glean/product-manager-ai-quality
- Sierra, Product Manager, Agent Development, Financial Services, via the AllthingsPM jobs catalog: /jobs/sierra/product-manager-agent-development-financial-services
- Deel, Product Manager job description template: https://www.deel.com/job-description-templates/product-manager/
- Indeed, Job description list by industry: https://www.indeed.com/hire/job-description-list
- AllthingsPM jobs catalog, 116 PM job descriptions at 18 AI companies, analysed 26 September 2026: /jobs
