Productboard

Senior AI PM Dominik Ilichman pulls back the curtain on how the team built Productboard Spark — AI designed to help product managers ship the right products, faster — and how PMs can make the most out of it. Watch the full webinar on demand → https://lnkd.in/d93-PRX5

We keep treating team speed like it's just 5x individual speed. It's not. One person moving at 10x while four people wait for context? That's actually slower. Speed only compounds when your tools are built for collaboration, not just output. Find out for yourself: https://lnkd.in/gWXU_97m

"Can't I just do this with Claude Code?" It's the first question people ask about Spark, our new PM agent. So I went straight to the source for an answer. 😄 Give Spark a try and see for yourself! 👇

Stop rebuilding context. You know the feeling: tab switching between tools. Copying feedback into an LLM. Re-explaining competitive intel to yet ANOTHER tool. Manually rebuilding the same product context. Again and again. What if you never left your working space? With Productboard Spark, PMs now have a specialized product agent that can bring together their scattered product data while helping them carry out their core product responsibilities. ✍️ Drafting a product brief? Pull feedback, roadmap context, and competitive analysis in one flow. One conversation. Full alignment. 🔎 Hunting for patterns in customer feedback? Surface opportunities and validate them against engagement metrics instantly. 🌎 Understanding your competitive landscape? Identify product gaps and learn why you may be losing users. With Spark, you're synthesizing context at scale, in one place. Use MCP connectors to query live data from tools like Amplitude, Linear, Pendo.io, Hex, Notion, and more via custom connectors. Use integrations to attach files from Confluence, Google Drive, and Notion to your Spark prompts and inform its outputs. Try Spark today → https://lnkd.in/drErv2zd

Happy Friday Product Managers! Repost: Denny Klisch 👉 https://lnkd.in/g-xDriiy

Users say they want AI to do tasks for them. That's not where the real value is. Jordan Nolff and the Productboard team just shared their lessons from building Spark, and one insight stood out: The real value isn't AI that completes tasks for users. It's AI that helps them think better. PMs don't need an agent to spit out a PRD. They need help thinking through customer needs, competitive alternatives, and business context. Other hard lessons from their journey: → They built the UI in 2 weeks, then spent 2 months iterating on AI quality. Most teams don't plan for this. → They started at 40% accuracy. Shipped at 85%. Ship early, and iterate with customers.  → Customers will accept 70-80% accuracy if they see you improving weekly. Perfection at launch matters less than visible progress. → Traditional estimation is broken. Frontend + backend ≠ done. AI quality is a third dimension nobody budgets for. The teams shipping great AI products aren't the ones with all the answers. They're the ones learning fastest from what didn't work. What's the biggest surprise you've encountered building AI products? Full article in the comments below 👇 #ProductManagement #AI #Startups #ProductDevelopment

Building AI products breaks all the normal rules. Jordan Nolff and our product team just wrote about what we learned building Spark on Ravi Mehta’s Substack — including the things that definitely didn't go according to plan and how we’ve evolved our approach. Turns out, there's a massive gap between "we can build this" and "this actually works well enough for customers to trust it." Some things that caught us off guard: → The timeline math is completely different. Build the UI in 2 weeks, spend 2 months on AI quality. Traditional project estimation doesn't work anymore. → Users lie to you (unintentionally). They'll tell you they want automation, but what they actually need is help thinking problems through. We had to reframe our entire approach. → 40% accuracy isn't the end, it's the beginning. The real question is: can you systematically get from 40% to 85%? We learned how (a labor of love). → You can't build for everyone at once. Early adopters will embrace imperfect AI. Your enterprise customers expect polish. For now, you have to pick one. Jordan shares his thoughts and brings together insights from Dominik Ilichman, Sr. AI PM for Spark, and our CEO Hubert Palan. Read it here → https://lnkd.in/dqt-_9qn