15 Oct 2026
How AI is changing user onboarding, one level at a time | Ramli John (Delight Path)
Most SaaS teams are adding AI to their onboarding right now, but few have a clear picture of what that should look like. In this episode of Love at First Try, the podcast about what makes products sticky and lovable, Jim sits down with Ramli John, founder of Delight Path, creator of Product Leaders Lab and author of Product-Led Onboarding and EUREKA. Ramli helps product teams turn more of their users into happy customers, and he has talked to dozens of product leaders about what's actually working with AI. We get into how to define taste and delight, why AI makes talking to your users more important than ever, the four levels of AI in onboarding he sees across the market, and the simple activation change that keeps paying off: just pick one.
🧠 What you'll learn in this episode:
0:00 - Ramli's story: Delight Path, Product Leaders Lab and two books on onboarding
1:54 - How to define taste: you learn it like food, by tasting a lot
3:42 - Delight means exceeding expectations, and what Uber got right
6:10 - Why AI is pushing product teams to talk to users more than ever
9:13 - AI is just the average of what's online, so human insight is your edge
10:45 - How to train your taste: study what you like and learn from your peers
13:52 - Why one audience's delight is another audience's annoyance
17:28 - EUREKA: define success first and design onboarding backwards
18:31 - The four levels of AI in onboarding: tells, guides, generates, acts
24:23 - AI agents and avatars that do the work for users: ready or cringe?
30:20 - An activation story: cutting a complex SEO tool down to 2-3 core jobs
33:01 - How to find your users' jobs to be done with 5-10 onboarding calls
35:17 - Ramli's AI workflows: Granola, Claude and ClickUp working together
39:52 - Ramli's favorite software and why Granola keeps winning
💡 Actionable takeaways from Ramli
Steal these quick wins:
- Define what success looks like, then design onboarding backwards from it. Starting from the end result keeps every step pointed at the moment users actually get value.
- Just pick one. Instead of a dashboard with ten things to do, route new users to the two or three core jobs they came for. Too many options make people freeze and leave, and focusing is the simple change that moves activation.
- Find those jobs in 5-10 customer onboarding calls. Ask why people signed up and why they bought. Ramli finds this beats staring at page views and events, because the calls tell you the motivation behind the clicks.
- Talk to your users more, not less, because of AI. A five-minute conversation can teach you more than a pile of surveys, and AI is essentially an average of what's online. The insights outside that average are what make your product stand out.
- Map your AI features to the four levels: AI tells (a chatbot on your knowledge base), AI guides (it launches the right tour or hotspot), AI generates (custom templates for each user) and AI acts (it does the task for the user after they approve). Each level is harder to build but gets users to value faster, so pick the level that fits your product and team.
- Design delight for your audience, not for everyone. What Notion users love can annoy busy users who just want to get the job done, enough to make them leave. Delight is exceeding the expectations of the people you serve.
- Train your taste on purpose. Analyze what you like and don't like in products and why, and spend time with peers. Like chefs learning from other chefs, you pick up taste from people whose judgment you trust.
- Automate the boring parts of your week. Ramli connects Granola to Claude through MCP to prep meetings, suggest weekly content from his calls and keep ClickUp updated, which frees his time for the conversations that matter.
