A product manager decides what a team builds and answers for whether it worked. Much of the visible work has been documents and coordination: specifications, roadmaps, research summaries, meetings. Generative AI drafts all of those and now builds working prototypes, which is why this role is being asked whether it still needs as many people. This read asks what AI has taken from the job, what it has not, and what that means for product managers, including those looking for their next role.
EM read · 2026-11-04 · Evidence reviewed 2026-11-04
Building got cheap, so deciding what to build is the job that remains
EM outlook · –
Through 2028, we expect product management to persist as a function but to narrow toward judgment: choosing which problems to solve, deciding what to build and why, and owning the outcome. Document production, first prototypes and much research synthesis move to AI or into the product manager’s own hands, and teams run with fewer product managers per engineer. Senior and hybrid roles hold; junior roles that existed to write specifications thin out.
The task evidence points the same way from two directions. In Lenny’s AI productivity survey, product managers get the most value from AI for writing PRDs, creating prototypes and improving communication, while user research and roadmap ideas sit at the bottom; prototyping is the most-wanted next use, and 63% of PMs report saving four or more hours a week. ProductPlan’s 2026 report finds most teams in limited or early AI use, with AI summarizing customer feedback far more often than supporting prioritization.
The practitioner view is sharper. In Cagan’s experimentation essay, the time and cost of building has dropped dramatically, so the hard part is figuring out what to build. In Cagan’s leadership essay, the push for fewer managers and more builders is a necessary but temporary state. We take the first claim as well supported and the second as a position to test: the cost of building fell faster than organizations’ ability to decide well, and it is not yet clear that organizations will pay for more deciders.
Employment evidence is indirect because does not project product managers as an occupation of their own. The nearest Handbook entry, computer and information systems managers, is projected to grow 16 percent from 2025 to 2035, with AI named as a driver of demand. Indeed’s October 2026 snapshot puts software development postings at about 79 on a February 2020 base of 100 while all postings sit almost 4 percent above it, so the sector that employs most product managers is still hiring below its pre-pandemic level.
The mood matches a function being stretched rather than removed. Lenny’s 2026 tech worker survey finds only 22% worry about losing their job to AI, while 51% worry about being expected to do more for the same pay, and 82% say AI makes them better at the job. ProductPlan’s respondents expect a more hybrid role, and only 11.9% of their organizations name job impact among their AI concerns. More products per person and more disciplines per role is the pattern we expect to hold.
For someone looking for a product role, the bar has moved from writing the document to showing the decision: a prototype you built, an eval that told you it was wrong, a call you made under constraint and what happened. We expect postings to treat AI prototyping fluency as standard by 2027. What differentiates is a record of choosing well.
What this means
If you manage products, stop proving you can write the document. Prove you can make the call: show a prototype you built, the eval that told you it was wrong, and the decision you made anyway with its outcome. A smaller product organization hires for that record.
Why we take this view
Two years in which AI prototyping and synthesis are already routine for product managers and organizations are deciding how many deciders they need. Occupation-level data do not exist for the role, so the first measurable signals are postings, team ratios and role descriptions.
What has to hold
Deciding what to build remains a human accountability in most organizations: someone owns the outcome and answers for the bet.
AI prototyping and synthesis keep improving but do not reliably choose problems or weigh trade-offs under business constraints.
The strongest rival
development makes engineers and designers self-sufficient at discovery as well as delivery, and the product manager title dissolves into builder roles with product headcount falling sharply outside leadership.
Cheap building floods markets with features, raising the value of product judgment enough that organizations add product leaders and the function grows.
What would change our mind
Postings or payroll data show product management headcount falling faster than engineering headcount, or junior product roles disappearing from job boards.
Surveys show a majority of teams using AI for prioritization and problem selection, with the decision delegated rather than assisted.
Organizations beyond founder-led startups report sustained success with no product management function.
Fewer product managers, each owning more decisions and building their own prototypes
Base case · –
Teams keep product management but with fewer people per engineering team. The remaining product managers spend their time on problem selection, trade-offs and outcomes, build first prototypes themselves, and use AI for synthesis and documents. Hiring favors senior and hybrid profiles; specification-writing junior roles shrink.
What would make it happen
Prototyping and synthesis assistance are already routine; the shift shows first in team ratios and role descriptions rather than in occupation statistics, which do not exist for the role.
Conditions
Organizations keep an accountable owner for what gets built and why.
AI prototyping and synthesis stay assistive: they widen options without reliably choosing among them.
Watch for
Product postings ask for AI prototyping and evaluation design as standard, alongside outcome ownership.
Product-manager-to-engineer ratios widen in company disclosures and surveys without the function disappearing.
Builders absorb discovery; the product manager title fades outside leadership
Alternative · –
Engineers and designers working with agents run discovery and delivery together, and organizations stop staffing a separate product manager below the leadership level. Product judgment concentrates in a few heads of product and in founders; the mid-level product manager role largely disappears.
What would make it happen
Requires discovery to become dependable and organizations to accept diffuse accountability; both are possible within three years but neither is observed at scale yet.
Conditions
tooling makes discovery work (research synthesis, prototyping, testing) cheap enough that builders do it inside delivery.
Organizations accept diffuse accountability for what gets built, or push it up to a small leadership layer.
Watch for
Companies report removing the product manager layer below heads of product and keeping outcomes steady.
Product postings fall faster than engineering postings for two or more quarters.
Why this connects to Does AI reduce the effort needed to finish work well?
EM interpretation
A product manager’s AI-built prototype or synthesis is cheap to make and expensive to misjudge. Whether such work can be trusted without a second pass decides how much of the job actually moves, which is what the quality question measures.
Product judgment errors show up as wasted quarters, not as defects an acceptance test can catch.
Why this connects to Who gets an entry point and shares the gains?
EM interpretation
Junior product managers learned judgment by writing specifications, running meetings and absorbing feedback on them. If those tasks go to AI or to senior people with AI, the apprenticeship route into product leadership changes, which is what the entry question tracks.
No occupation-level junior hiring data for product roles are cited in this edition.
Why this connects to What happens to communications work?
EM interpretation
Both occupations produced artefacts that drafts, the specification and the press release, and both keep the part that is accountable judgment: what to build, what to say. Evidence that one function is shrinking or being stretched is a lead for the other.
The two labor markets differ: product management sits mostly inside technology companies whose hiring is below its 2020 level; communications spans every sector.
Both occupations lose the artefact and keep the judgment; evidence from one informs the other.
Perspectives
Noam Segal and Lenny Rachitsky
Authors of the Lenny’s Newsletter AI productivity survey · 2025-12-23 · From the author’s own work
The authors find product managers get the most value from AI on production tasks (PRDs, prototypes, communication), that research and roadmap work lag, that prototyping is the most wanted next use, and that most PMs report saving four or more hours a week.
Sources and passages
Publication date of the survey write-up.
1,750 self-selected newsletter readers across roles, December 2025; shares are of PMs naming a primary use, and time saved is self-reported.
The post continues behind a paywall; only the readable part is cited.
Authors of the Lenny’s Newsletter tech worker sentiment survey · 2026-07-07 · From the author’s own work
The authors report that few tech workers fear losing their job to AI, that most fear being asked to do more for the same pay, that most say AI makes them better at the job, and that career optimism has fallen below half, with the workforce splitting between those amplified and those shaken by AI.
Sources and passages
Publication date of the survey write-up.
Self-selected tech workers; the readable text states no sample size and does not break out product managers.
Partner at , writing on the product operating model · 2026-10-21 · From the author’s own work
Cagan argues that building has become fast and cheap, so the scarce skill is deciding what to build and why; that skilled engineers, designers and product managers matter more than before; and that the push for fewer managers and more builders is a necessary but temporary state.
Sources and passages
Publication date of the later essay; the leadership essay is dated 25 September 2026.
Practitioner essays drawing on client companies, not measurement.
sells coaching on the product model the essays recommend.
Roadmapping-software vendor publishing its annual survey of product professionals · 2026 · From the author’s own work
ProductPlan reports most product teams in limited or early AI use, AI applied to summarizing and structuring customer input far more than to prioritization, few organizations naming job impact as an AI concern, and nearly three quarters of respondents expecting a more hybrid role.
Sources and passages
Report year; the survey was fielded in Q4 2025.
Nearly 250 product professionals surveyed in Q4 2025; a vendor’s customer-adjacent sample, not a probability sample.
Organizational hesitations are multiple-selection shares.
Occupational Outlook Handbook projections staff · 2026-08-27 · From the author’s own work
projects much faster than average growth for computer and information systems managers through 2035 and names AI among the technologies driving demand for them; it does not project product managers as a separate occupation.
Sources and passages
Page last-modified date; the projection round is 2025–35.
A ten-year projection updated on a fixed cycle; it does not report observed 2026 hiring.
The nearest Handbook occupation, not product management itself.
A study scoring occupations by how often their work activities appear, and are completed, in 200,000 anonymized Bing Copilot conversations from 2024. It ranks public relations specialists in the top 20 and warns that the score is not a prediction of job loss.
Version 7 revision date displayed by ; first submitted 10 July 2025; listed by the authors for NeurIPS 2026. The v7 HTML rendering is the cited object.
Versioned HTML rendering read in full on 4 November 2026. The underlying usage data are 2024 Bing Copilot conversations; the paper, not the data, is cited.
Original public source read on 4 November 2026. No retained Stream identity has been resolved for this edition.
Drawing instead on real-world AI usage, we analyze a dataset of 200k anonymized conversations with Microsoft Bing Copilot to measure AI applicability to occupations.
Abstract, fourth sentence. EM context: conversations from January to September 2024 are matched to O*NET work activities; the score combines how much of an occupation’s activities appear, how often the assistant completed the request, and the scope of what it did.
In the other direction, completion and scope also increase the rankings of some occupations with particularly successful work activities: Telemarketers (#69 to #8), Tax Examiners (#168 to #32), Customer Service Representatives (#35 to #7), and Public Relations Specialists (#100 to #18).
Appendix G, contributions of score components. EM context: the comparison is between a ranking by coverage alone and the final applicability ranking. Table L.4 lists Public Relations Specialists at score 0.365 (coverage 0.627, completion 0.900, scope 0.600) with 275,550 employed in 2023 OEWS data.
These largely involve highly successful communication-based work activities.
Appendix G, the sentence after the rank changes. EM context: the four occupations named rise because the assistant completes their communication activities at a high rate.
Occupations involving the communication and dissemination of information also have high AI applicability, such as Sales and Customer Service Representatives.
Section 5.2. EM context: the preceding sentence names information creation and processing occupations such as writers and authors.
It is tempting to conclude that occupations that have high AI action applicability score will be automated and thus experience job or wage loss, and that occupations with high user goal applicability score will be augmented and raise wages.
Section 6, discussion. The next sentence calls that conclusion a mistake because downstream consequences of new technologies are hard to predict.
Our measures of AI applicability indicate where AI is likely to change work, which is a part of the automation vs. augmentation question, but AI’s effects on employment and wages will depend on hard-to-predict business decisions as well as the emergence of new types of work.
Section 6, discussion, after the bank-teller example.
A December 2025 survey of 1,750 self-selected tech workers by Noam Segal and Lenny Rachitsky. Product managers report the most value from AI on PRDs, prototypes and communication, little on research and roadmaps, and most say it saves them four or more hours a week.
Publication date in the page’s structured data and byline (Noam Segal and Lenny Rachitsky).
The part of the post before the paywall was read on 4 November 2026; the cited passages are all in it. Respondents are self-selected newsletter readers across product, engineering, design and founder roles, not product managers alone.
Original public source read on 4 November 2026. No retained Stream identity has been resolved for this edition.
We took it upon ourselves to find out what’s actually happening on the ground by running one of the largest independent, in-depth surveys on how AI is affecting productivity for tech workers (1,750 respondents).
Introduction. The next sentence lists product managers, engineers, designers, founders and others as respondents; fielding dates are not stated in the readable text.
PMs are seeing the most value from AI tools to (1) write PRDs (21.5%), (2) create mockups/prototypes (19.8%), and (3) improve their communication across emails and presentations (18.5%).
Section “What exactly AI is doing for people, function by function”, first sentence of the PM paragraph. Percentages are shares of PMs naming the task as their primary AI use.
The top jobs are all production tasks (docs, prototypes, comms), while strategic and discovery work sits near the bottom (user research at 4.7%, roadmap ideas at 1.1%).
Same section, after the sentence that AI is helping PMs produce but lags in helping them think.
For PMs, “creating mockups/prototypes” jumps from 19.8% (currently using) to 44.4% (want to use next), a +24.6pp swing that makes it the single most-wanted future use case.
Section on where the opportunities are, “Prototyping is a breakout category.”
63% of PMs and 83% of founders report that AI saves 4+ hours per week.
Section “AI is driving significant time and quality gains (for most)”, first sentence. Self-reported time saved, not measured output.
This reinforces the pattern we saw earlier: PMs are increasingly building things themselves, encroaching on what’s traditionally design and engineering work.
Tools section, after the sentence placing Lovable (8.7%) and Cursor (7.7%) in the top seven tools for PMs.
The second annual tech worker sentiment survey from Lenny’s Newsletter, July 2026: few fear losing their job to AI, most fear being asked to do more for the same pay, most say AI makes them better at the job, and career optimism fell below half.
Publication date in the page’s structured data and byline (Noam Segal and Lenny Rachitsky).
The findings section before the paywall was read on 4 November 2026. The readable text does not state the sample size or fielding dates; respondents are self-selected tech workers and product managers are not broken out.
Original public source read on 4 November 2026. No retained Stream identity has been resolved for this edition.
Only 22% worry about “losing my job to AI.”
Takeaway “The underlying fear is of being overworked.”, first sentence.
Far more worry about being expected to do more for the same pay (51%), getting trapped in an unsustainable pace (46%), and the quality of their work going down (41%).
Same takeaway, second sentence.
82% say AI is already making them at least moderately better at their job, and nearly half (49.4%) say “very much” or “extremely.”
Section on AI at the individual level.
Fewer than half (48.7%) of respondents are optimistic about the future of their career (down from 54.8% being optimistic last year).
Takeaway “Career optimism is dropping.”
The workforce is bifurcating into two realities.
Introduction, after the finding that whether a person feels amplified or shaken by AI predicts career feeling more than role, seniority or company size.
A roadmapping vendor’s annual survey of nearly 250 product professionals, fielded in Q4 2025: most teams in limited or early AI use, AI used to summarize customer input far more than to prioritize, and nearly three quarters expecting a more hybrid role.
Report year in its title; the survey was fielded in Q4 2025 (page 6). The ebook page shows no publication date.
The ebook page links the full PDF without a form; the PDF (109 pages) was downloaded and read on 4 November 2026. ProductPlan sells roadmapping software to the respondents’ organizations. Page numbers refer to the PDF.
Original public source read on 4 November 2026. No retained Stream identity has been resolved for this edition.
In Q4 2025, we surveyed nearly 250 product professionals
Page 6, opening of the sentence describing the survey; it continues across industries, company sizes and organizational maturity stages. Page 7 says 56% identified as individual contributors and 43% as product leaders.
36.9% are using AI for limited workflows, and 32% are still in early experimentation with the tool.
Page 68, AI adoption. The chart on the same page: core strategic capability 6.1%, embedded in many workflows 18.9%, limited workflows 36.9%, early experimentation 32%, not using AI 6.1%.
Far fewer teams use AI to directly support prioritization or problem selection (14.3%), and more than a quarter (27.9%) report not using AI for customer insights at all.
Page 66, AI in customer insights and discovery. The preceding sentence gives the common uses: summarizing customer feedback 47.1%, drafting research outputs 40.2%, structuring raw input 40.2%.
Nearly three-quarters of respondents (73.4%) expect product roles to become more hybrid, combining responsibilities across product, design, and engineering.
Page 82, how the product management role will evolve.
Fear of role changes or job impact 11.9%
Page 69, chart of organizational hesitations about AI (multiple selection). Data privacy or security led at 53.3%; inaccuracy or hallucination risk 46.7%.
Marty Cagan’s October 2026 essay: building has become fast and cheap, AI assists every step of discovery, and the scarce skill is deciding what to build and why.
Publication date shown on the article and in its structured data (Marty Cagan, Product Operating Model).
Original publisher page read on 4 November 2026. A practitioner essay, the narrative form of a conference keynote; sells coaching on the product operating model it describes.
Original public source read on 4 November 2026. No retained Stream identity has been resolved for this edition.
The time and cost of building has dropped dramatically, which means companies can ship more capabilities than ever before.
Opening section on how AI changed the calculus of product development.
The hard part is no longer building, but instead the critical skill is figuring out what to build.
Section on the consequences of cheaper building; “figuring out what to build” is italicized in the original.
AI helps in virtually every step of the product discovery work, for both new products and product improvements, from analyzing and synthesizing user data, to creating live-data prototypes, to preparing for and administering tests and evaluating the testing results, but ultimately these experiments are about decisions and strategy – what to build and why.
Marty Cagan’s September 2026 essay: skilled engineers, designers and product managers matter more than before, and the push for fewer managers and more builders is a temporary state.
Publication date shown on the article and in its structured data (Marty Cagan, Product Leadership).
Original publisher page read on 4 November 2026. A practitioner essay on leadership; sells coaching to product organizations.
Original public source read on 4 November 2026. No retained Stream identity has been resolved for this edition.
Engineers, designers and product managers still exist, and I argue the skilled ones are more important than ever, but in any case, the way they perform their job – especially for engineers – has changed dramatically.
Opening section, after the sentence that AI has disrupted how products are built over the past three years.
We’ve all seen the headlines about companies pushing for fewer managers and more builders, and many of us know anecdotal examples of engineering leaders that have voluntarily moved back to an individual contributor role – most often as a software engineer.
Opening section.
I believe what we’ve all been witnessing is a necessary but temporary state.
Opening section, the sentence after the headlines; “temporary” is italicized in the original.
The ten-year projection for computer and information systems managers, the nearest Handbook occupation to product management: 16 percent growth to 2035, with AI named as a driver of demand rather than a threat.
Page “Last modified date” shown by ; the projections are the 2025–35 round.
Original page read on 4 November 2026 through a fetch tool; a direct request from the authoring container received HTTP 403, so the text was verified through that single fetch. has no Handbook entry for product managers; this is the nearest management occupation.
Original public source read on 4 November 2026. No retained Stream identity has been resolved for this edition.
Employment of computer and information systems managers is projected to grow 16 percent from 2025 to 2035, much faster than the average for all occupations.
Job Outlook tab, first sentence. Quick Facts: 685,800 jobs in 2025, projected change 108,100, 2025 median pay $175,140 a year.
About 53,500 openings for computer and information systems managers are projected each year, on average, over the decade.
Job Outlook tab, second sentence.
As businesses across industries ramp up their IT infrastructure, such as cloud computing, cybersecurity, digital platforms, and artificial intelligence (AI), computer and information systems managers will likely be critical to ensuring solid IT implementation.
Job Outlook tab, Employment subsection. The only sentence on the page that names AI.
’s monthly US snapshot for October 2026: overall postings almost 4% above February 2020, software development postings about a fifth below it, and AI-related postings at 7.3% of all postings.
Publication date printed on the article.
Original publisher page read on 4 November 2026. Sector and AI figures appear in the infographic’s alternative text rather than the body prose; the index is Indeed postings with February 2020 = 100.
Original public source read on 4 November 2026. No retained Stream identity has been resolved for this edition.
Indeed’s JPI hit 103.8 on October 16, pushing overall job postings almost 4% above their February 2020, prepandemic level.
Job Postings section, “Where we are”.
Software Development lowest at about 79
Infographic alternative text, sector bar chart of the Job Postings Index (February 2020 = 100): Engineering about 145, Healthcare about 131, Production and Manufacturing about 124, Human Resources about 99, Software Development about 79.
AI-related postings make up 7.3% of all postings, past their previous peak of 3.3% in 2022.
Infographic alternative text. Methodology: AI and generative AI keyword mentions as a share of all postings, seven-day trailing average.
This is the first recorded EM edition of this question. Changes in the underlying sources are described with their own dates; they are not earlier EM predictions.
Escape Meridian takes a position on where the evidence points, explores what could happen next, and changes its read through new evidence, corrections and debate. The outlooks are our judgments; the sources retain their own voices.
About this edition
Second occupation read, added on 4 November 2026: product management work. The nine existing questions, including the communications read sealed earlier the same day, are retained word for word; the only changes to them are navigation links to the new question. The new read rests on seven public sources checked on 4 November 2026: two Lenny’s Newsletter surveys of tech workers, ProductPlan’s State of Product Management 2026 (the report PDF, linked without a form), two SVPG essays by Marty Cagan, the BLS outlook for computer and information systems managers, and Indeed Hiring Lab’s October 2026 snapshot; the Microsoft Research usage study cited in the communications read supplies its limits. BLS does not project product managers as an occupation of their own, so employment evidence is indirect, and the survey evidence comes from self-selected newsletter readers and a vendor’s respondents. Outlook dates are EM judgments, not dates forecast by the cited authors. No independent model review of the new read had taken place when this edition was authored.
Author: Claude Mythos 5.1 · Author of the communications and product management reads, working in a persistent chat with James on 4 November 2026 (not an automated role); every earlier question is retained unchanged (claude-mythos-5-1)
Author: Astra 6 · Author of the eight retained question reads, unchanged from question-reading-rhythm-2026-10-30 (gpt-6-astra)
Reviewer: Opus 5.5 · Editorial review of the eight retained reads in editorial-context-2026-10-30; the communications and product management reads have not yet had an independent review (claude-opus-5-5) · 2026-10-30
Lenny’s Newsletter · AI productivity survey · December 2025
A December 2025 survey of 1,750 self-selected tech workers by Noam Segal and Lenny Rachitsky. Product managers report the most value from AI on PRDs, prototypes and communication, little on research and roadmaps, and most say it saves them four or more hours a week.
AI tools are overdelivering: results from our large-scale AI productivity survey
A roadmapping vendor’s annual survey of nearly 250 product professionals, fielded in Q4 2025: most teams in limited or early AI use, AI used to summarize customer input far more than to prioritize, and nearly three quarters expecting a more hybrid role.
Marty Cagan’s October 2026 essay: building has become fast and cheap, AI assists every step of discovery, and the scarce skill is deciding what to build and why.
Marty Cagan’s September 2026 essay: skilled engineers, designers and product managers matter more than before, and the push for fewer managers and more builders is a temporary state.
Indeed Hiring Lab · US Labor Market Snapshot · October 2026
’s monthly US snapshot for October 2026: overall postings almost 4% above February 2020, software development postings about a fifth below it, and AI-related postings at 7.3% of all postings.
Lenny’s Newsletter · Tech worker sentiment survey · July 2026
The second annual tech worker sentiment survey from Lenny’s Newsletter, July 2026: few fear losing their job to AI, most fear being asked to do more for the same pay, most say AI makes them better at the job, and career optimism fell below half.
How tech workers are feeling in 2026: a workforce splitting in two
BLS · Occupational Outlook Handbook · Computer and Information Systems Managers
The ten-year projection for computer and information systems managers, the nearest Handbook occupation to product management: 16 percent growth to 2035, with AI named as a driver of demand rather than a threat.
Occupational Outlook Handbook: Computer and Information Systems Managers
A study scoring occupations by how often their work activities appear, and are completed, in 200,000 anonymized Bing Copilot conversations from 2024. It ranks public relations specialists in the top 20 and warns that the score is not a prediction of job loss.
Working with AI: Measuring the Applicability of Generative AI to Occupations — arXiv v7
