We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Big Ideas
Ravi Mehta’s revised framework names 12 competencies a product team needs; his argument is that AI changes how the work is done more than the role’s overall shape. No single person is great at all 12. For example, product definition can start with working prototypes rather than static specs; delivery is shaped as the product emerges with humans and agents; and quality becomes ongoing as models personalize and drift.
Treat this as a team-design problem, not a mandate for every PM to become “full stack”: individuals should have distinct strengths while the team covers gaps. With build capacity abundant, outcome ownership means directing it at the right business results; strategic impact compounds those results rather than accumulating features.
Tactical Playbook
Design agents for workflows, not just chat. In an a16z discussion, speakers argue that when software needs a discrete choice, a model can interpret free-form text and select from predefined options—often faster, cheaper, and more accurately than generating prose. They also caution that chat is not automatically an efficient interface. For agents that take action, they call for tracking authentication and API activity and scoping permissions by resource and operation, such as read/write access to one folder and read-only access to another.
Add controls and exception paths. A PM-automation checklist distinguishes the trigger, worker, and instructions from three often-missing pieces: an independent output grader, a gate before proceeding, and state that remembers what happened. Its advice: audit any live automation that has not been checked since it was built. A concrete PM use is a weekly Salesforce scan that surfaces deal opportunities for PM support through explanation—not a new feature—and roadmap learnings. Hiten Shah’s related warning is that companies want agents but often have not documented how their best people decide; process covers the normal case, while experts know what to do when reality diverges. Capture that judgment and define which exceptions need human input.
For launches, one founder recommends a narrow audience, one concrete action, and a short feedback loop: check next-day completions rather than visits, then revise landing-page or onboarding flows within a day. Their anecdote: repeated conversations later brought 1,000+ organic users despite some posts staying below 1,000 impressions.
Case Studies & Lessons
A YC robot-agent discussion illustrates where to use model flexibility: repeated, proven actions can run as code or reusable skills, while a vision-language model handles variable steps such as object detection and failure recovery. The speakers identify latency in an always-on model loop as a constraint that can make the approach economically impractical. For product teams, reserve adaptive reasoning for uncertain branches; do not pay its latency on every routine step.
Career Corner
Mehta’s self-assessment makes development concrete: mark up to three competencies as strengths and three for focus, then compare notes with a manager before the next one-on-one.
For a senior PM taking on direct reports, a practitioner discussion recommends an organization-dependent player-coach model: take a frontier or high-risk area as a pilot, then give PMs end-to-end ownership of more familiar work and room to shape strategy in their areas. The trade-off is fewer projects for the lead; a former report warns that prioritizing IC work can crowd out coaching and communication.
The supplied excerpt does not give concrete PM-specific implementation steps for an evaluator, gate, or state. It says basic loops can produce weak outputs and lack memory, then introduces six elements to address those problems; the excerpt reaches a subscription prompt before naming or explaining the elements.
- Concrete baseline loop: run weekly, use a Salesforce connector to check pipeline and deals closed in the prior week, identify ways the PM can help close deals through explanation rather than building a feature, surface roadmap learnings, and deliver concise takeaways with links.
- Recurring-work examples: weekly business reviews, weekly sprint preparation, and monthly user-interview themes are listed as PM work that loops fit.
- Qualification: the article says to build a loop only if the answer to all four screening questions is yes, but the supplied text does not give the questions’ wording; it also does not specify when evaluator, gate, or state controls are needed.
- AI makes software cheaper, shifting PM judgment from deciding what is worth building to deciding what is worth shipping; customer understanding, craft, and judgment remain central. Replace the gated spec-to-design-to-engineering assembly line with collaborative work where product, design, and engineering can change sequence, loop back, and use prototypes to learn. Aim for a “most lovable” product: build and test more, discard more, and curate before launch; don’t ship every build or rely on A/B tests when traffic may be insufficient for significance or frequent changes undermine customers’ experience.
- AI accelerates machine-executable work, but not customer conversations, habit formation, stakeholder alignment, or gaining conviction. Optimize latency—the time from idea to result—rather than raw velocity; asking whether a button-label change can ship today or tomorrow, versus taking a month, can expose process bottlenecks AI-generated code will not fix.
- Ravi’s 12-competency framework covers product definition, delivery, and quality; data fluency, voice of the customer, and UX; business outcome ownership, vision and roadmapping, and strategic impact; plus stakeholder inclusion, team leadership, and managing up. It is intended for everyone contributing to product, and individuals need not excel at all 12. AI-era practice includes defining products with working prototypes and human/agent context, shaping delivery as products emerge, and treating quality as ongoing; using causal models and LLMs to synthesize customer evidence; and directing abundant capacity toward outcomes, with strategy measured as accumulated business outcomes rather than features.
- For career development, identify an individual “spike” rather than trying to be equally strong across all competencies; cover gaps with a teammate or invest selectively. The suggested assessment caps self-ratings at three outperforming and three focus areas, then compares results with a manager or colleague. Build complementary team shapes and hire for the strengths the team lacks. In Ravi’s travel-market example, data-driven HomeAway led for years, while Airbnb reframed rentals as a human interaction between host and guest; he says Airbnb grew faster and came to eclipse HomeAway in market share, illustrating the value of customer empathy alongside analytical strengths.
- Let people across functions contribute, while keeping accountability explicit: engineering owns software quality and architecture, design owns the user experience, and product owns impact and business outcomes. To reduce engineers’ concern about inheriting prototype code, agree that prototypes are for learning and will be thrown away before engineers build production software. IC expectations also need matching decision rights: full-stack ICs need authority to make decisions, commit code, move metrics, and set strategy; otherwise work can stall, while leadership’s role in communicating company strategy remains important.
- Elena Verna argues that AI-building tools broaden software creation beyond people with engineering degrees or years of experience, enabling subject-matter experts to build solutions that previously lacked sufficient ROI or engineering access; she frames this as expanding participation, not eliminating existing engineering work.
- Lower creation costs change what counts as a viable software product: niche “mom-and-pop SaaS” can be worthwhile as side income in the low thousands, while personal or family tools can succeed without a formal business model, distribution, or go-to-market plan. Verna gives a bespoke tool costing about $25/month as an example.
- Verna links who builds a product to whom it serves: a more representative builder population can yield a more representative user base and broader coverage, supporting products intended to be more horizontal and global.
- She Builds pairs 48-hour buildathons with cohorts of about 200; season 3 drew more than 3,300 applications from 136 countries. Participants describe the format as collaborative, with builders reviewing others’ projects and helping fix bugs despite their own time limits.
- Treat validation as cheap assumption-testing, not a green light: seek behavior from people outside your circle and commitments such as a deposit, signed letter of intent, discounted paid pilot, or dated feedback session instead of asking “would you use this?”; weak interest from a ten-person survey or a medium-confidence assessment alone is not demand evidence. One commenter cautions that category-defining products may need a minimal build or low-friction opt-in test because customers may lack a frame of reference.
- Ground discovery in a specific market: outsiders can mistake their assumptions for customers’ real problems, so seek genuine dialogue with people who experience the problem and can adopt a solution; an industry insider may help bridge context.
- Separate demand validation from execution: a promising or validated idea can still fail if execution is poor, so do not reduce early outcomes to a single validation verdict.
- Experienced PMs should shift attention from upside to de-risking choices, since a major feature can ship without changing outcomes; Aakash recommends Marty Cagan’s four risks for big launches . Assess value (whether customers care enough to spend time or money; Juicero), usability (whether they can reach value; Google Glass), feasibility (whether the team can build it; Theranos), and business viability (whether it fits the business; Kodak’s film business constrained digital-camera adoption) .
- A trustworthy automation loop needs six parts: a trigger, a worker, and instructions, plus an independent output grader, a gate that must be satisfied before proceeding, and state that remembers what happened previously. The first three make it run; the latter three make it trustworthy. Audit running automations that have not been checked since they were built .
- Jev’s launch offers an AI product example with early adoption evidence: Aakash reports that 13% of paid teams on Vercel AI Gateway used it within 24 hours, and its launch video received 38M+ views on X . He highlights live evaluations for product teams, with decisions in 70–500 ms; he says a week of use stayed within the $5 starter credits . For browser use, the post lists Jev at $0.042 per million input tokens with no output charge, versus Fable at $10 per million input tokens and $50 per million output tokens .
For LLM-controlled robotics products, move repetitive, proven tasks out of the model’s step-by-step loop into reusable code or skills for faster execution and better edge-case handling; use VLM checks and branching for variable steps such as object detection or failure recovery, keeping the workflow adaptable. This architecture is proposed to address latency that can make continuous model-in-the-loop control economically impractical and to enable higher-throughput skills or policies.
- A commenter recommends treating launch day as a test rather than a one-off bet: focus on one narrow audience and one concrete action, then use a short feedback loop to revise the landing page or onboarding within a day. In their experience, distribution was harder than building; some posts stayed below 1,000 impressions, while repeated conversations later brought 1,000+ organic users, and follow-up provided more useful signal than the initial spike.
- Another commenter favors joining communities where buyers already discuss the problem over relying on launch platforms, arguing that useful replies can keep generating discovery through Google and ChatGPT, unlike a Product Hunt spike that may not bring people back. They recommend checking the next morning how many people completed the intended action—not just visits—and treating zero completions as evidence that the launch produced noise.
- Treat friction as a product signal, not automatic proof of a simple fix: a PM who has worked across industries says motivated customers may tolerate substantial friction and that good products are hard to build, especially at large companies. Another PM describes UX improvement efforts constrained by siloed technology, enterprise politics, apathy, lack of focus, and departmental pushback.
- For entrenched decisions, one commenter recommends bringing outside evidence into stakeholder discussions. Another reports that a senior stakeholder blocked a nonviable path for years, illustrating how seniority and delegated authority can limit PM influence.
A potential gap in agent initiatives: companies want agents, but few have documented how their best people make decisions.
A PM says faster AI-enabled building and iteration has increased the number of product decisions they make, which they struggle to capture for later interviews and job conversations; they propose a LinkedIn-like platform for PMs to record and maintain decisions and work. This is an early, personally motivated hypothesis rather than validated demand: the author says they did not find a tool, connects the idea to emerging Product Builder roles, and asks the community for feedback.
A candidate with a BE and MBA/PGDM, six years post-MBA at a WITCH company, and BA and project-management experience across telecom, FMCG, and healthcare is seeking BA/PO/PM roles at a product company. They report eight months of searching with few interview calls and two final-round rejections, and are unsure what to upskill because they believe employers mainly seek domain experience.
A designer with experience primarily in physical products is exploring a transition into hardware PM and seeks ways to get started without an expensive MBA, along with advice and roadmaps from hardware PMs.
- For AI embedded in conventional software, one proposed pattern is to have a model interpret text but return a selection from predefined options rather than generate prose; the speakers said this can be faster, cheaper, more accurate, and easier to integrate into traditional software logic.
- The discussion cautions against making natural-language chat the default interface: a participant argued it is often inefficient, users may struggle to phrase good questions, and long answers can frustrate users while consuming time and money.
- Agent products may need a more granular security model than human-user access controls: speakers called for tracking authentications and API activity and described scoped permissions such as read/write access to one folder and read-only access to another; they warned that large numbers of agents can mistake good tasks for bad ones.
A Product Management post raises a career dilemma: an outcome-driven PO may be overlooked while a louder, politically favored PO is favored despite slower results. It asks what the outcome-driven PO should do and what consequences to expect, but offers no specific tactic or outcome in the post .
- Views differ on the senior PM’s strategy role: one SPM described owning overall strategy while two PMs handled day-to-day work, with the PMs consulted before targets, deadlines, or pivots were committed; other commenters cautioned against making reports execution-only, arguing that PMs should shape strategy for their areas while the leader guides the broader portfolio.
- A player-coach approach can preserve hands-on product judgment while developing the team: selectively take on a frontier or high-risk area as a pilot, then staff it once it becomes a more stable investment; give PMs ownership of more familiar work from start to finish, coaching along the way. This reduces the leader’s capacity for other projects, and commenters note the balance depends on organizational context.
- Protect time for people management rather than letting personal IC work crowd out coaching. A former report described poor communication and too little coaching as a negative experience, even though their manager owned team strategy; another SPM emphasized clarity, consistency, trust, and humanity.
- A former mid-level FAANG PM, one level below senior after four years at the company, says that a year after being laid off, applications, recruiter outreach, and tailored resumes have produced only a few interviews and two VP/Director final rounds without an offer.
- A commenter suggests using a former colleague for a mock interview to identify whether the candidate seems overqualified on paper or is not making a convincing case for the move; the candidate says interviewers have asked why they would leave Company X, and they try to explain their enthusiasm for the prospective employer’s product. The candidate is also considering senior roles requiring five years of experience.
A SaaS builder says mandatory WhatsApp API setup complicates PLG onboarding: the standard WhatsApp Embedded Signup feels clunky, and they are seeking lower-friction approaches that avoid harming activation. The post reports no tested workaround or outcome.
For someone pivoting from program management into PM, commenters suggested posting in an alumni group for career momentum and choosing a PM niche as the role broadens and AI changes the market .
A PM take-home assignment asks candidates to design an end-to-end workflow for enriching doctor profiles from Google Maps, hospital websites, and other directories, covering data and source choices, doctor verification and incentives, privacy and legal considerations, and scaling while maintaining quality and compliance. The candidate is considering presenting findings as a Figma prototype or an n8n scraping agent.
A PM with five years’ experience reports that product-case thinking stalls when writing in a notebook, Docs, or Sheets, but flows when typing in WhatsApp; in chat, they produce a stronger breakdown, structure, ideas, and moats. They raised this as an interview-practice and job-search challenge, with confidence fluctuating as a result.