# AI changes Nothing Dax Raad, OpenCode | AI Engineer CODE 2025 | 16:39 Source: https://www.youtube.com/watch?v=o3gmwzo-Mik Channel: AI Engineer (https://www.youtube.com/@aiDotEngineer). Summarised by AIE Talks. Page: https://aietalks.com/talks/ai-changes-nothing Published: 2025-11-23 Tags: design, product-strategy ## TL;DR - AI does not solve the human work of attracting users, getting them to an aha moment, and retaining them. - Marketing works when an idea is cool, funny, deep, or surprising enough that people want to share it. - Long-term retention comes from building powerful primitives underneath a simple first experience, which takes product understanding and sustained effort. ## Summary Dax Raad argues that AI changes how software can be built, but it does not remove the hard parts of making a product succeed. Products still need marketing that earns attention, a clear aha moment that gets users to understand the value quickly, and a design that keeps advanced users from outgrowing the product. Raad says AI has not produced good ideas for him in marketing or product design. It tends to generate corny ideas, and it cannot discover the right mental model or primitives for an unfamiliar problem space. His advice is demanding: try ideas that might be shared, remove every unnecessary step before the aha moment, and build broad, powerful primitives before assembling a simple experience. He is also honest that OpenCode does not perfectly follow this advice. Understanding the principles is easier than executing them every day. ## Key ideas ### Product success still depends on three human moments [01:32](https://www.youtube.com/watch?v=o3gmwzo-Mik&t=92s) Raad frames product growth as a funnel that takes someone from knowing nothing about a product to hearing about it, trying it, liking it, and fully converting. He focuses on three moments: attracting attention through marketing, getting the user to a singular aha moment, and retaining the user over time. These problems have been part of his whole career, and AI has not made them much easier. The funnel framing also connects the moments. Marketing effort is wasted if only a small share of people reaches the product's value, and a product cannot grow well if it keeps losing users after they start using it. ### Marketing needs ideas people want to share [02:08](https://www.youtube.com/watch?v=o3gmwzo-Mik&t=128s) Raad says marketing begins with stopping people who are busy with their own lives and getting them to think about your product. A feature blog post, an influencer discussing a release, or a billboard may explain what was built, but these efforts rarely make someone want to show it to friends or coworkers. The better target is an idea that feels crazy, funny, deep, or surprising enough to prompt, "Can you believe that they did this?" He compares marketing to swinging at products: teams will miss often, but they should take shots that could become hits. AI has not given him a good idea and tends to make brainstorming "super corny." ### The aha moment requires ruthless removal of friction [06:41](https://www.youtube.com/watch?v=o3gmwzo-Mik&t=401s) After earning attention, a product has to get users to the moment when its value clicks. Raad calls this the singular aha moment, where someone understands what the product is and feels sold on it. Teams often want to expose users to many features, but he says they must choose the one interaction that matters most and deprioritize everything else. Every step before that moment should be examined and cut. He criticizes signup flows that ask about company size, employee count, or job title before users know whether they want the product. If only 3% of people reach the aha moment, the work spent attracting them has largely been wasted. ### A product needs a clear reason to exist [08:49](https://www.youtube.com/watch?v=o3gmwzo-Mik&t=529s) Raad says some products may not have an aha moment at all, which should force the team to reconsider what it is building and whether the effort is worthwhile. He uses ChatGPT's input box as an example of a powerful first interaction: users can type anything and receive a humanlike response, so the product is understandable immediately. He also points to Facebook's poke, which let someone flirt with a crush while keeping the action deniable. These examples are extreme, but his standard is that every product should have one moment that communicates, "This is the reason I exist." Finding it requires positioning, clarity, and a distinct understanding of the product's space. ### Retention means serving simple users and demanding power users [10:50](https://www.youtube.com/watch?v=o3gmwzo-Mik&t=650s) Once users reach the aha moment, the product has to give them reasons to stay for life. Raad focuses on people who have used the product for a long time and are pushing its limits. They want more configuration, advanced features, and support for specific team or enterprise needs. The difficulty is adding that capability without damaging the simple first experience that helped new users understand the product. Open source makes this tension especially clear. A hobbyist may want something simple and free, while a large company may need to host it themselves and control sensitive data. ### Power comes from primitives underneath a simple experience [12:41](https://www.youtube.com/watch?v=o3gmwzo-Mik&t=761s) Raad rejects the idea that a product must choose between being simple and being capable. He says teams often call an incapable product simple, but a product can offer both when the underlying design is strong. Instead of building a feature directly, he recommends creating broad, advanced primitives that can be assembled into the simple experience used by most people. More advanced users can later access those primitives directly and bend the product to their needs. This approach takes more work at the start, but it reduces the chance that users will outgrow the product and move to something else. ### AI cannot discover the right primitives without deep understanding [14:14](https://www.youtube.com/watch?v=o3gmwzo-Mik&t=854s) Building primitives requires first understanding the problem space well enough to know which primitives are needed. Raad says AI can help with conversation or brainstorming, but it cannot replace that exploration. The team must communicate a clear understanding of what it wants before AI can contribute meaningfully, and reaching that understanding takes time and energy. This is why he does not expect AI to make people suddenly excellent at product design. The hard part is forming the mental model that lets a team design a system that can handle many future situations. ### Knowing the advice does not make execution easy [14:55](https://www.youtube.com/watch?v=o3gmwzo-Mik&t=895s) Raad warns that OpenCode itself gives users plenty to criticize, even though he can explain the product principles behind the talk. The advice is simple to understand and very difficult to apply every day. Getting marketing, onboarding, product design, and retention to work together takes sustained effort from a person or team. AI lets people do things they could not do before, but it does not remove the daily pain of trying to make something great. Raad's conclusion is deliberately unsentimental: his life is still as hard as it has ever been, and making something amazing still demands the same work. ## Notable quotes - "You can't do cool." (06:23) - "You need to figure out what is the singular aha moment in your product and eliminate friction ruthlessly to get someone to that aha moment as fast as possible." (07:02) - "You can build both a simple and capable experience." (13:00) - "AI is amazing and lets us do all kinds of things that we couldn't do before. Does not save us from that day-to-day pain in trying to make something that's great." (15:35) ## Tools & references mentioned - OpenCode - ChatGPT - Facebook - SST ## Who should watch - You are building an AI product and expect the model to solve your marketing or onboarding problem. - Your signup flow exposes many features and asks for information before users understand the product. - You need a product that can start simply while giving experienced users enough depth to keep using it. ## Related talks - [Your AI Product Will Fail Unless You Can Explain It](https://aietalks.com/talks/your-ai-product-will-fail-unless-you-can-explain-it) (Veronica Hylak, Hey AI, 06:02) - [Good design hasn't changed with AI](https://aietalks.com/talks/good-design-hasnt-changed-with-ai) (John Pham, SF Compute, 20:25) - [Everything is ugly, so go build something that isn't](https://aietalks.com/talks/everything-is-ugly-so-go-build-something-that-isnt) (Raiza Martin, Huxe, 25:15) - [The #1 Question Every AI Product Manager Must Answer](https://aietalks.com/talks/the-1-question-every-ai-product-manager-must-answer) (Brian Balfour, 00:56) - [Why Your Product Needs an AI Product Manager, and Why It Should Be You](https://aietalks.com/talks/why-your-product-needs-an-ai-product-manager-and-why-it-should-be-you) (James Lowe, Incubator for AI, UK Government, 18:37)