Sneha Keshwani

for the past year, I have seen too many startups falling into the trap of "shipping the AI PRD." let me explain: a PRD (product requirement document) is what a PM writes to communicate their vision to engineering and design teams. there are startups building AI that helps PMs chat and generate a PRD faster. here's the problem: that's a transient bottleneck. because tools like Lovable are skipping the PRD entirely. you don't need a document when you can show and communicate way better. when you're building with AI, you have to avoid building the "AI version of the old process." you have to ask: how do i circumvent this entire process and rip it apart at its seams? that's the long-lasting strong company. examples of AI PRDs to avoid: - AI that makes meetings more efficient (vs removing meetings) - AI that speeds up handoffs (vs removing handoffs) - AI that writes better docs (vs removing docs) the question isn't "how do we make the old way faster?" the question is "what becomes possible when we remove the constraint entirely?" that's how you find the real opportunity.

17

2 Comments

I've been lucky to advise startups and mentor PMs on AI strategy and product careers over the past few years - it's one of the most rewarding things I do. As these conversations have grown, I'm adding a bit more structure so I can keep showing up well for everyone. Going forward, please use https://lnkd.in/gYX5dVe7 to book time directly. If you're a founder building in AI, a PM navigating your next career move, or just thinking through AI in enterprise, I'd love to chat! Looking forward to more of these conversations, not fewer. :)

89

1 Comment

New Article: 7 Patterns That Derail Product Teams This week’s post is for PMs, engineering, stakeholders, and even aesthetic-only UXers who have ever wondered why a pro UXer “pushes back” in meetings. I break down 7 real-world patterns that cause miscommunication, and show how to get aligned on the goal, the constraints, and what counts as evidence so your team can move faster without solving the wrong problem. I hope some teams find this one helpful. https://lnkd.in/gwDHGHbV

5

2 Comments

Startup CEOs: With very few exceptions, the first PM you "hire" at your company should be an engineer or designer who's already on the team. I wrote the article below a decade ago, and this is 10x more true today than it was then, due to the product, design and engineering roles increasingly converging. Article: https://lnkd.in/gXYN5ETj

166

9 Comments

The PM's job is now to identify the handful of true non-negotiables and let the rest go. That's the argument Cat Wu, Head of Product for Claude Code at Anthropic, made on Lenny's Podcast. At Anthropic, roles are blending together — designers ship code, engineers make product decisions, PMs build prototypes and evals. She named product consistency as the main tradeoff. This is compelling, and probably true for their context. It's also the kind of argument that's easy for executive leaders to hear and walk away with the wrong takeaway: that blurring roles is the path to velocity for any team. Anthropic is an AI-native company with small, elite teams operating on a compressed feedback cycle where a wrong bet costs an afternoon. Their users have high tolerance for rough edges because they understand the product is moving fast. That context does not describe most product organizations. The product team I lead owns a restaurant back-office solution covering food cost, labor, inventory, and compliance across hundreds of brands and tens of thousands of locations. Product consistency is key in this space and is a meaningful tradeoff when roles blur. Cat's not wrong. But it's not the most important thing that breaks. The bigger risk is the accumulated user context that doesn't live in a codebase and can't be reverse-engineered from usage data. A developer with strong product instincts still doesn't know that operators count inventory at 6am in a freezer with no internet access, that shift managers have about three minutes between rushes to look at anything, or that a UI change affecting scheduling can ripple across an entire brand's labor cost model. Someone has to own that context, because without a dedicated discipline, it's the first thing to go when the pace picks up. The "let the rest go" sentiment also assumes the cost of getting it wrong is a quick rollback. At enterprise scale, a rushed release touching payroll or inventory means operators discovering errors after a pay cycle has run, or making decisions on an incorrect cost of goods. Speed still matters. But in this domain, a wrong bet doesn't cost an afternoon. Cat's world moves at a pace few product organizations can match, and the model they've built is impressive. The danger is when that becomes the template. Anthropic's Claude Code team is a specific context: AI-native, small team, building for developers. Assuming that model transfers to enterprise B2B, regulated industries, or any domain where operational context is load-bearing is how teams end up shipping more and solving less. There's a lot worth learning from Anthropic, but it matters whether your product, your users, and your context are anything like theirs. Lenny Rachitsky Catherine Wu https://lnkd.in/eGhDk8nY

11

1 Comment

Shipping the right feature at the wrong time.... is still the wrong feature. Early in my PM journey, I thought a “great product” meant shipping the right feature with great UX & engineering. But users don’t reward perfection. They reward urgency. If they’re not already duct-taping a fix with spreadsheets, Slack threads, or hacks; then it’s probably not hurting enough yet. Now my rule is simple: Don’t just solve what’s right. Solve what’s right now. Prioritize the pain that's already real. Everything else can wait. Curious if you have experienced this?

40

1 Comment

If you’re looking for real startup experience — not just a title, but actual ownership on a meaningful product— this is worth your attention!👇🏼

1

If you know the solution before you define the problem, you aren't writing a PRD. You're writing a story. So I built my personal PRD Copilot - https://lnkd.in/g9ACK9Ax It’s not just a template filler. It acts as both a critic for your drafts and a brainstorming partner that forces you to think problem-first and build from scratch. It works in 3 modes - 1. 𝗕𝘂𝗶𝗹𝗱 𝗳𝗿𝗼𝗺 𝘀𝗰𝗿𝗮𝘁𝗰𝗵: Start with a raw idea. It asks the clarifying questions you 𝘴𝘩𝘰𝘶𝘭𝘥 be asking yourself, nudging you to validate the "why" before the "what." 2. 𝗦𝘁𝗿𝗲𝘀𝘀-𝘁𝗲𝘀𝘁 𝗲𝘅𝗶𝘀𝘁𝗶𝗻𝗴 𝗱𝗼𝗰𝘀: Drop in a half-baked doc, brain dump, or "final" PRD. It critiques it through multiple lenses: Is the problem actually clear? Would engineering & design buy into this? Are you solving a user need, or just shipping activity? 3. 𝗣𝗿𝗼𝗯𝗹𝗲𝗺-𝗦𝘁𝗮𝘁𝗲𝗺𝗲𝗻𝘁 𝗠𝗼𝗱𝗲: Throw in a vague, uncomfortable problem. It helps you shape a feature idea grounded in that problem - stripping away your solution bias. Why I built this 👇 Because I wanted someone who would pressure-test my thinking and help me drive clarity - ensuring my features don't fail in execution. If you’re a PM who wants lesser fluff and more conviction in your PRDs - give it a try. #CustomGPT #PRDMaker #PRDCopilot #productdoc #PM

182

26 Comments

Week 1 of the Product Management Fellowship forced me to slow down. Most of my work so far has been in early-stage startups with no standard Product Management processes, with urgency, ambiguity, and figuring things out as we went. That environment pushed execution. It didn’t leave much room to think before building. Going back to basics of product thinking, CVP, and user journeys helped me recognize patterns I’ve already lived through, but never paused to name. Two ideas stayed with me:  1. Product thinking is about restraint    In early-stage startups, the default question is usually, “Can we build this?”    week-1 reinforced why it’s more useful to ask:     • What were users doing before this existed?     • What pain points were they already tolerating?     • What new problems could this create after launch?    Most of the rework I’ve seen didn’t come from poor execution. It came from skipping these questions.  2. Second-order effects matter more than the first fix  The "WhatsApp message delete" trace example shows how even small features need to account for trust, context, and downstream behavior, not just the immediate use case. Looking back, stronger fundamental thinking would have saved time, rework, and confusion in several projects I’ve worked on. That’s the gap I’m working on now, bringing more structure into how I think before committing to solutions. For those who’ve moved from chaotic startup environments into more structured product orgs which fundamentals helped you reduce rework the most ? Thanks to Akhil Yash Tiwari and the Product Space for these insightful sessions.

20

If you're in product ops you'll immediately recognize this dilemma. A massive copy-paste challenge to maintain a commercialization/marketing view of the world...and a product view of the world. Made 300% harder because someone wants a "simple calendar" of launches. The reality is: 1. Initiatives have a M:1 relationship with Launches 2. Different launch tiers have different launch milestones 3. Calendars must include milestones, launch, and initiatives 4. Initiatives can have multiple releases 5. ...but not all releases are launch related 6. Oh, there are massive internal "launches" that marketing doesn't care about And this is why spreadsheets prevail. God-awful spreadsheets with: 1. Dozens of fields (like 6 dates for milestones) 2. Random manual fields to list multiple initiatives 3. Terrible VLOOKUP date madness 4. Product themes that are distinct from anything product uses A classic example of how there is a "right" model, but the UX complexity kills everyone, so they opt for something crappy. (PS: I love these problems. Hit me up if this resonates)

43

8 Comments

I wrote about why product managers should step into the shoes of their customers and dispassionately figure out if they would buy their own product if they were a customer https://lnkd.in/gZDzuVFY

25

3 Comments

Excited about today's PM Brief as it highlights why I’m really enjoying this project... 1️⃣ Teresa Torres wrote a fantastic summary of how LLMs work. If you’ve been curious but a little intimidated, it’s worth slowing down and really digesting it. 2️⃣ I pulled an insight from her piece and framed it around something directly actionable: how PMs should technically understand their own products. This is exactly the kind of benefit I want The PM Brief to deliver. Hope it resonates with you as well! Links in the comments... #ProductManagement

25

3 Comments

In my early Product Management journey, I once thought success hinged on crafting standout features. Over time, I've realized it's more about enabling exceptional outcomes, even when not directly involved in every task. My growth trajectory includes: - Transitioning from "What should we build?" to "What issue are we addressing?" - Shifting from solo work to empowering my team's abilities - Moving from rigid roadmaps to focusing on concrete results Product Management goes beyond execution; it's about orchestration. True triumph arises when your team excels autonomously, free from constant supervision. What pivotal mindset shifts have shaped your Product Management voyage

5

🧩 Week - 6: From Build Completion to Strategic Realignment Last week marked a shift, from Phase 1 execution to planning what comes next, with tighter alignment and sharper structure. 🧠 Presented our MVP (Cavo) alongside team leads to the founder, walking through the product’s core system flow, design intent, and key differentiators 📅 With Phase 1 complete, I reworked the Product Roadmap to reflect delivery realities and rollout logic → The updated version is now prepped to release next week, with execution tied to five clearly structured phases 🔗 Scoped and mapped task dependencies across frontend, API, model, and prompt teams; setting up parallel execution without bottlenecks 👥 Finalized team structure: → Assigned pod leads, clarified reporting lines, and locked ownership flows → Made sure every pod knows who approves, who executes, and who escalates 🚀 Helped new PMs get started by walking them through the system, team workflows, and live pods 🧱 Stepped in to unblock backend and model teams by clarifying data flows and logic expectations tied to our architecture 📊 Team outcome: • Phase 1 closed cleanly • Roadmap now phase-aligned with real delivery flow • Ownership structure live • Onboarding friction nearly eliminated ⚡ What teams usually underestimate at this stage: 📉 61% of roadmap failures happen after Phase 1, due to lack of structural recalibration 🧭 57% of new contributors lose 3��5 days without onboarding systems 🧩 Execution clarity ≠ velocity — it creates velocity (Source: Notion, ProductPlan, First Round Review) We’re not moving into Phase 2 with momentum. We’re moving in with precision. #ProductManagement #AIProductManagement #OpenforFTRoles2026 #ExecutionClarity #InternshipExperience #RoadmapDesign #TeamAlignment #CrossFunctionalWork #ZeroBudgetExecution #StartupPM #BuildWithIntent #RecruiterReady

5

🚀 The 5-Hour PRD: Streamlining Product Requirements in the Age of AI Fellow PMs: how long does it take you to move from a fuzzy idea to a validated PRD and wireframes? I recently tested an AI-driven product workflow that did it in just 5 hours - and the results were eye-opening. Here’s how it went 👇 🧩 My AI-Powered Product Workflow (End-to-End in 5 Hours) 1️⃣ Define the Audience & Problem I have defined detailed personas for my Notion AI Agent: David (CISO), Rachel (Security Engineer) and James (Risk Manager). Then I asked Notion AI to analyze our current reports (all the knowledge is in Notion), identify the key data gaps, and outline the specific questions each persona would ask. After a few iterations and scoping, I had a clear, persona-driven problem statement and out of scope. 2️⃣ Move from Text to Visuals Next, I switched to Google Stitch [https://lnkd.in/d7cmSc7c], gave it the problem statement as a prompt and included a couple of screenshots from our Figma design library for inspiration. With a few structured prompts, I generated three high fidelity wireframes: dashboard overview, stats view, and object-level drill-down. 3️⃣ Validate with AI I dropped those wireframes back into Notion and asked AI: “Do these visuals answer the personas’ questions?” It gave a detailed, point-by-point validation: and the alignment was spot on. 4️⃣ Finalize the PRD With validation complete, I asked the agent to write the high-level solution and detailed user stories describing interactions (think: “click this, then that happens”). All in: five hours, start to finish. ⚙️ The Bottleneck This speed is incredible: I could almost hand the PRD and wireframes to an AI dev tool like Claude Code/Codex and say “build it.” But here’s the problem: engineering velocity hasn’t caught up. If PMs can deliver fully fleshed PRDs in hours, yet development still takes weeks, the bottleneck has shifted. 🤔 The Question for PMs When product can move this fast, how do we enable engineering to keep up: leveraging AI, automation, and smarter workflows to match product’s new pace? Would love to hear how other PMs are tackling this balance between AI-accelerated product design and AI-assisted engineering execution.

32

7 Comments

I just went through the hardest interview process of my career while interviewing for the Director of Product role here at Subject. It’s a funny thing to interview for a new role at a company you already work at, especially when you’re being evaluated by a team that already knows exactly what you’re capable of and where you fall short. Most interviews let you present a polished version of yourself. You can highlight your strengths, downplay your weaknesses, and control the narrative. But that’s not the case when you're interviewing with colleagues who’ve worked alongside you for months. They knew my blind spots and had seen me struggle with certain problems. Naturally, they were curious about the areas where they’d seen me grow and struggle, because they genuinely wanted to understand how I’d approach leadership with that experience. I spent sleepless nights the week before, preparing insights that would surprise people. I created a one-page data sheet showing four key metrics I thought the company should focus on. The intensity of internal interviews was unlike anything I’d experienced, but it was also the most rewarding, because the feedback came from people who genuinely cared about the company’s success and my growth.

14

The hardest job in product isn’t shipping features. It’s building trust. Trust with engineers that you won’t waste their time. Trust with designers that their work won’t be watered down. Trust with leadership that you’ll deliver outcomes. Trust is the invisible currency that makes teams move. How do you build trust in your team? Connect with me for career advice, product management, or resume review: https://lnkd.in/g6_HbgqA

9

3 Comments

Scaling doesn’t reveal how mature a product is. It usually reveals how much velocity was required to compensate for misalignment. There’s plenty of good advice about moving beyond feature delivery, being more intentional, clearer about outcomes, and more deliberate about how product work is framed. I see these as directionally right, but incomplete. In earlier product environments, progress often meant moving fast in the direction of whoever had the strongest pull. Work started quickly, and value was often obvious enough to justify after shipping. At larger scale, that stops working. Not because teams lose judgment, but because when many capable teams move in parallel, misplaced effort compounds fast, and the cost of being wrong becomes explicit. What changes most isn’t problem complexity, it’s that the value must be articulated before execution, not discovered through it. At a certain point, even expert intuition ceases to function as a coordination mechanism. Not because expertise disappears, but because intuition doesn’t scale across dozens of decision surfaces. That’s where maturity truly becomes evident. In more developed product environments, alignment isn’t seen as overhead: it’s regarded as essential product work, encompassing framing, shared language, explicit trade-offs, and clarity on what not to do. A somewhat unsettling conclusion I keep returning to: velocity looks like progress... until the product has enough blast radius for misalignment to matter. Photo by Ahmad Odeh on Unsplash.

16

My next guest, Eno (co-founder of Factory), built this insane product management AI skill that blew my mind. It includes: → Product principles and positioning → What an 11-star experience looks like → PRD templates and review rubrics → Prioritization frameworks ...and more It's 700+ words and I'll admit it made me a little worried about my profession 😅 Watch the full interview now: https://lnkd.in/gapCAhf4

279

10 Comments