AI Workflows PMs Use Every Day

The agent is just the tip of the iceberg. An impressive AI demo is easy. Building a workflow you can rely on every day is hard. Reliability lives in the plumbing: the right context, connected tools, maintained instructions, deterministic steps, human checks, and visibility when something breaks. IMPORTANT: Complete this form prior to the event to receive your entry pass from Cribl via email. On September 30, ProductTank SF is bringing together product people to share the AI workflows they rely on every day. Not prototypes, not polished demos, but the systems that are crucial to how they work. Each speaker will show their workflow in action, unpack what makes it reliable, and give you a practical path to build on it yourself. KEY TAKEAWAYS Real workflows — tested systems speakers use in their day-to-day work Under the hood — the context, tools, instructions, documentation, and human judgment that keep them running Build your own — examples, resources, and a shared GitHub repo so you can experiment afterward THE FORMAT Short talks → Audience Q&A Each speaker gets 10–12 minutes to show their workflow in action, followed by 10–15 minutes of audience Q&A to dig deeper into the details. THE SPEAKERS Jorge Alcántara, CEO and Co-Founder of Zentrik.ai, will show how he combines Playwright, Storybook, and Codex to automate product video creation. Albert Collell, Senior AI Product Leader at Qurrent and product management instructor at Stanford, will show how an LLM-powered personal wiki turns meetings and interviews into a searchable second brain. Tom Alterman, Director of Product at GoodLeap and founder of Tomek.io, will show how he uses AI to take a product idea from concept to pull request. More speakers to be announced. Attendance will be first-come, first-served. We will close doors once we reach capacity; arrive early to secure a spot. Special thanks to our generous sponsor: Cribl: Simplify how you route, transform, and reduce data with a central tier that provides choice, flexibility and control.