We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Big Ideas
Trust is a product capability, not a synonym for privacy. Scott Belsky distinguishes privacy—keeping data private—from trust, which requires understanding an agent’s judgment and reasoning and being able to audit or inspect it. A Mind the Product speaker argues that agents need explicit context for what “good” means: vision, strategy, goals, and principles, with “trust over short-term gain” as a foundation; without that context, they default to average outputs, slop, or hallucinations. Translate this into product requirements: authorization and identity, explicit consent and context, fallback and kill-switch mechanisms, and logs of inputs, outputs, and intermediate actions. The practical design pattern is human control without deskilling: in radiology, the doctor diagnoses first and AI flags disagreement as a safety check; in government hiring, excluding proxies such as ZIP code and commute time is treated as a core product requirement.
Safe change is becoming the pricing advantage. ZoomInfo’s Henry Schuck says customers who tried consumption pricing saw their AI bills and panicked, while outcome-based pricing is difficult when many go-to-market steps separate software usage from a closed deal. Hiten Shah’s takeaway is sharper: SaaS pricing may never settle, so the durable advantage is the ability to change it safely.
PM’s future skill stack looks more builder-like. Aakash Gupta quotes Freshworks CPO Srinivasan Raghavan predicting that Engineering, Design, and Product Management will converge into “Product Builders”—a forecast, not a settled job-market fact. The actionable progression is AI fundamentals, prompt and context engineering, tool fluency, and a data-first operating system; the same checklist emphasizes prototyping to a screen, grading it with evaluations, learning agent distribution, and recognizing that shipping is only one-third of the job.
Tactical Playbook
Discovery: ask what the problem has to beat. A customer confirming pain proves that a problem exists, not that it deserves action now. Map what consumes their time, what they already pay to fix, and where the problem ranks; if it is near the top, investigate the workaround and next commitment, and if it ranks low, treat it as deferred value. Ask what they have already tried: extensive attempts followed by disappointment with existing solutions are stronger demand evidence. Replace “Would you use this?” with “What would this displace, and what have you already tried?”
Protect execution with a shared evidence trail. One product lead describes a politically exposed, multi-business-unit program with dependencies, an accelerated timeline, and poor documentation, where an escalation questioned their ability only three weeks into the role. The practical countermeasure is simple: document decisions and risks, align expectations early, and maintain shared records of constraints and progress.
Case Studies & Lessons
Vertical integration helps when sequence is the value—but early-stage economics can be punishing. YC’s founder stack combines deck sharing, pitch-meeting scheduling, SAFE distribution, and related workflows to simplify the founder experience and feed data back into the system. Product Hunt’s Ship applied the same logic to landing pages, pre-launch email collection, surveys, targeted updates, distribution, and re-engagement; several thousand founders and companies used it. But 5–20% of its early-stage projects shut down monthly, while limited willingness to pay among makers and startups capped revenue and larger enterprises could manage best-in-class tools themselves. Integrate tightly around handoffs that create learning or re-engagement, then stress-test churn and willingness to pay before expanding into a broad platform.
Career Corner
Show shipped impact, not just tenure. One hiring manager says domain expertise mattered, but selected a smart-home/IoT PM who had launched a freemium app and completed the conversion-optimization cycle despite having half as many years of experience as competing candidates. Another current recommendation is to demonstrate data-driven prioritization, customer impact and measurement, lightweight Codex or Claude Code prototypes, and practical agent fluency—not feature shipping alone. Build portfolio stories around a shipped decision, its outcome, and the trade-off behind it.
Tools & Resources
Use “screen, then grade” as a weekly AI-PM exercise. Gupta’s resource path links AI-PM practice, prototyping and evaluations, agent distribution, and PM/Team/Company operating systems; its concrete advice is to get to a screen, grade it, and build a first eval. Pick one workflow, define what good looks like, test it, and inspect failures before adding complexity.
- Treat the first enterprise customer call as an intelligence-gathering session, not a pitch: keep it informal, let the prospect speak first, ask open questions about organizational change, and avoid recording, slides, or a demo; use the answers to shape the eventual product narrative.
- Co-design enterprise demos with the internal champion in a 15-minute preparation call covering the agenda, features, likely questions, and attendees. Show only the roughly 20% of the product relevant to the account, allowing prospects to pull in additional capabilities as needed.
- Use a tightly scoped validation pilot: select three or four power users, assign specific tasks, define success together, and limit the test to 48–72 hours. Extend to a one- or two-month paid pilot only when deep integration is required, with the fee credited toward the contract if the customer proceeds.
- Capture feedback immediately after the demo: obtain a raw champion debrief within five minutes, identify where stakeholders disengaged and who could block the deal, and treat a previously active champion going quiet as a signal that internal conditions have changed.
The enterprise sales cycle is framed as 15 stages rather than the commonly assumed five, with an explicit warning against skipping steps. The discussion covers a “pincer model” for securing the first meeting, a 2–3-sentence cold-outreach pitch, an intro call designed to extract maximum intelligence, a 2–3-day pilot structure, and tactics for navigating pricing and procurement—useful GTM context for PMs working on enterprise products.
- Define the outcome before optimizing the headline metric. The startup claims up to 48 hours’ earthquake notice and more than 80% accuracy from public California predictions. The discussion challenges that metric unless it specifies the meaningful time horizon, location or epicenter precision, magnitude tolerance, and treatment of false positives and false negatives; frequent minor events can otherwise make a high score commercially irrelevant.
- Make trust a product requirement for high-stakes AI. Publish predictions and historical results users can audit, expose false positives and false negatives, backtest on real data, and benchmark against simple baselines; the advice explicitly calls for verifiable research, peer review, and validation on real-world data. A commenter who said they read the paper countered that it used off-the-shelf ML trained on historical Greek earthquake data rather than a revolutionary advance, reinforcing the need for independent technical review before breakthrough positioning.
- Sequence go-to-market around buyers with an immediate economic consequence. The advice identifies local and state agencies—including emergency response, utilities, education, and transportation—as likely first buyers, followed by risk-exposed sectors such as insurance, investments, and manufacturing; grants may help fund early R&D. Public predictions, social media, and PR can open conversations, but should not substitute for evidence; agency risk managers, emergency responders, and scientists will scrutinize the methods and technology before major contracts or funding.
Enterprise selling is framed as a 15-stage process rather than the commonly assumed five, with the discussion covering a “pincer model” for landing the first meeting, 2–3-sentence cold outreach, intelligence-focused intro calls, a 2–3-day pilot structure, and pricing/procurement navigation. JJEN Abel characterizes the broader problem as “90% of companies are doing sales wrong.”
- Pricing strategy: SaaS pricing may remain volatile, so the durable product advantage is the ability to change pricing safely rather than waiting for the market to settle. Consumption pricing can create unpredictable bills for customers, while outcome-based pricing is difficult when many go-to-market steps separate software usage from a closed deal.
- Controlled testing method: Segment customers before testing and limit the initial experiment to one region, segment, or industry; ZoomInfo is testing a pre-bought consumption model with a subset of customers. Run the test for 2–3 months, gather comprehensive customer feedback, and use direct customer evidence rather than rapidly changing external commentary to form a pricing point of view.
- Rollout and incentive safeguards: Expand a pricing model to another region or industry when feedback is positive; roll it back and iterate when it is not, limiting learning costs and avoiding damage across the full customer base. Provide quota relief during tests by crediting sales representatives as if smaller experimental deals were full-sized, so rep compensation does not bias the test against the new model.
- An “agentic coding” workflow would run eight parallel tabs on new features in the same repository and have agents detect overlapping work, manage worktrees, merge changes, run tests, and resolve Git conflicts autonomously, with a human adjudicating substantive conflicts afterward.
- Adapt uncertainty communication to the listener, not the speaker. People process uncertainty through different filters—quantification, certainty/optimism, or worst-case anxiety—so the same message can land very differently; leaders should identify their own filter before communicating and distinguish intended delivery from what the recipient actually hears.
- For product and team leaders, more data or reassurance can backfire: a data-oriented founder may overwhelm a nervous team with dashboards, while an optimistic manager may dismiss reorganization concerns by saying everything is fine. Listen for the recipient’s fear and realign with it before adding explanations or trying to persuade.
- Build evidence of shipped product impact, not just tenure. Hiring feedback favors domain fit plus proof of building and launching a real product; one manager hired a smart-home/IoT PM who had launched a freemium app and completed the full conversion-optimization cycle despite having half as many years of experience as competing candidates.
- Demonstrate full-lifecycle, outcome-oriented PM craft. Interviewers look for outcomes and metrics, strategy as well as execution, and experience spanning ideation, launch, optimization, and sunset. Prepare detailed stories showing stakeholder influence, team leadership under pressure, mistakes, and what you learned.
- Tailor positioning to the target role rather than seeking universal appeal. Domain expertise may be decisive for some hiring managers but only a tiebreaker for others; when changing domains, emphasize business outcomes and ways you have empowered product teams, while newcomers can add value by questioning entrenched assumptions. A design background can be a differentiator in design-centric organizations when framed around effective collaboration with designers.
- Use data, customer impact, and hands-on AI fluency as differentiators. One commenter recommends leading with data-driven prioritization, defining feature success by customer impact and measurement rather than shipping alone, and demonstrating lightweight prototypes built with tools such as Codex or Claude Code plus practical knowledge of agents.
- Enterprise sales should be treated as a 15-stage process rather than a five-stage funnel; skipping a stage is presented as potentially deal-killing.
- The playbook includes a “pincer model” for securing the first meeting, a 2–3 sentence cold-outreach pitch, intro calls focused on extracting intelligence, a structured 2–3 day pilot, and deliberate handling of pricing and procurement.
- Avoid running a single unprepared demo: the source calls this the worst sales-process mistake and says it can lose the deal.
- AI product teams are using subsidized access to generate improvement data: Thinkymachines made Inkling free on OpenRouter for several weeks, limited access to agentic harnesses, and plans to use account-disassociated usage data to improve its agentic performance.
- Julie Zhuo expects this pattern to become more common, with labs offering otherwise valuable products or services for free to collect data that improves models; she highlights their strong appetite and financial capacity to fund better data.
- Trustworthiness by design should be treated as a core product requirement for AI-enabled products, not a compliance add-on: make the product’s vision, strategy, goals, and principles explicit, with “trust over short-term gain” as a governing principle. This is especially important when agents participate across discovery, analysis, development, and scaling, because unclear context can lead them toward unintended, low-quality outcomes.
- Operationalize the framework by defining measurable trust outcomes and explicit agent rules around human autonomy, harm prevention, fairness, explicability, transparency, non-bias, privacy, and societal benefit. Implement supporting controls such as authorization and agent identity, explicit consent and context, fallback and kill-switch mechanisms, and logs that capture inputs, outputs, and intermediate actions so results are reproducible.
- The talk gives two concrete patterns: a radiology workflow keeps the doctor responsible for the diagnosis while AI flags disagreements as a safety check, helping avoid skill atrophy; a government hiring system treats excluding proxies such as ZIP code, commute time, and résumé gaps from the model as a core product requirement to demonstrate fair decisions. Preserving user agency can also be a growth lever: the speaker says Ecosia gained 40% more U.S. users after allowing people to use search without AI summaries, while Google did not offer that choice.
- Validation-oriented product design: The founder says the team converted a seismologist cofounder’s peer-reviewed algorithm into a usable service that forecasts earthquakes up to 48 hours ahead, with California predictions and historical data published for public checking. The founder reports 80% accuracy and 83% recall for the California demo, while acknowledging missed events, false positives, and lower accuracy for forecasts made years ahead.
- Trust and evidence before scaling: A commenter recommends making the methodology clear, repeatable, transparent, and rigorously data-validated so independent scientists could reproduce it; another suggests validating the process, sharing it with researchers, using their endorsements to win initial customers, and starting in one geographic region before scaling. The founder plans blockchain verification to lock predictions before outcomes, while acknowledging that the historical track record is not proven to have been posted in advance.
- Narrowing the initial market: The founder’s focus shifted from selling to governments toward precision-manufacturing infrastructure such as semiconductors, rail networks, and dams, where forecasts could support operational safeguards.
Startup and product work should not be treated as formulaic: use experience to recognize patterns, but assess the specific situation clearly before deciding on the next step.
When ideas decay quickly, speed itself becomes a form of research—making rapid iteration part of the learning process.
- PMM’s strategic value is company-dependent: practitioners describe high-functioning teams being involved early by product and sales to do positioning work, with some PMM organizations also shaping product strategy and business priorities; other companies use the PMM label for content, campaign, events, or collateral execution.
- Suggested path from content/comms into PMM: pursue an internal pivot by telling your manager about the goal and taking stretch projects with the PMM team; build a portfolio by completing PMM-style case studies or projects in your current role.
- AI increases the value of evidence-based, cross-functional work: commenters warn that copywriting- and content/collateral-heavy PMM roles are vulnerable to AI, while recommending customer research, positioning, launches, sales support, data collection, customer conversations, and competitor intelligence as higher-value capabilities.
- Hiring-market caveat: one practitioner reports that PMM hiring can be highly exact-match—favoring six-plus years in a specific business model—and recommends internal promotion as the most viable route; this is an individual account, not a market-wide benchmark.
- Hiten Shah’s principle for product decision-making: use patterns as guidance and frameworks to save time, but when reality presents an unfamiliar situation, rely on judgment rather than forcing a formula; he says this perspective comes from more than 20 years of building companies.
- A PM who is the only team member outside an offshore PM team says they feel sidelined on a shared product, receiving “leftovers” and struggling to engage effectively.
- The initial diagnostic questions focus on clarifying the operating model: the PM’s role and reporting line, the offshore team’s structure and location, what “shared product” and “leftovers” mean, how work is divided, and whether other product teams use the same setup.
- A 2026 graduate reported landing an off-campus APM role at a top cloud company with an 18 LPA offer and advised aspiring PMs not to pursue paid courses.
- Community advice for entry-level PM preparation: avoid paid cohorts that repackage freely available frameworks and templated case studies; instead build curiosity through product teardowns and conversations with users. The commenter argued that repetitive case studies may hurt candidates because hiring managers see them repeatedly.
A builder seeking a non-technical collaborator for customer interviews, adoption diagnosis, onboarding and activation experiments, landing-page and positioning tests, and pricing/packaging experiments was told that this work resembles a product manager role with an emphasis on UX research. A practical engagement path proposed was to find a skilled UX/usability researcher through a trusted referral, start with a short-term contract, pay hourly or per project to control costs, and avoid taking on a partner or paying with equity.
- Validate the procurement workflow, not just category interest: Before building a two-sided marketplace, narrow discovery to one material and one buyer type; interview procurement teams about their last real purchase and the criteria that mattered—price, specifications, certification, lead time, and supplier risk. Talk to 15 European buyers that already have a green-materials mandate and map how they currently find suppliers; brokers, trade shows, and ERP vendors may be the real alternatives.
- Prove transaction behavior manually: Broker the first few deals before building the marketplace—suggestions range from three to ten—and test whether buyers and suppliers will complete the process with founder assistance. If they will not, software is unlikely to solve the underlying problem.
- Separate stated value from switching and monetization: Buyers may value lower-carbon materials yet resist changing vendors, compliance workflows, or payment terms, so validate switching behavior rather than interest in another marketplace. A post-deal take rate was suggested as easier to test than charging both sides subscriptions before the marketplace has liquidity.
