Web News: AI vs No-Code

The episode centers on an article by Chris McCauley titled 'Why We Tore Down Our No-Code Site and Went Back to Code,' which serves as a jumping-off point for a broader discussion about whether AI-powered agentic coding workflows are replacing traditional no-code tools. Mike opens by arguing that a transition is already underway, where developers and agencies are increasingly using LLM-powered tools to build and iterate on websites faster than WYSIWYG no-code platforms allow, particularly for marketing sites that require rapid A/B testing and content iteration.

Matt provides a grounding counterpoint, noting that many small-to-medium businesses are not choosing their web platform based on technical capability — they often call and request Webflow or WordPress by name because their teams are familiar with those tools. He also highlights that a large portion of his clients barely have time to fill in the blanks of a pre-built CMS, let alone manage an agentic coding workflow. He argues that the 'rambunctious, tech-savvy entrepreneur' demographic is a flashier but smaller slice of the market, and that most small businesses don't care if their footer color hasn't changed in five years.

Mike pushes back by explaining that agentic tools are becoming increasingly end-to-end — handling hosting via partnerships with Vercel and Netlify, compressing and optimizing images automatically, running web vitals checks, and tightening the client feedback loop from weeks to days. He argues that even for agency owners like Matt, the agentic workflow would free up time by automating content research, competitive analysis, and iterative changes, allowing them to serve more clients with less overhead.

The conversation expands into a more philosophical question raised by Matt: if AI agents are writing all the code, why would they use React or Next.js at all instead of compiling directly to vanilla HTML, CSS, and JavaScript — or even lower-level outputs? Mike concedes this is a valid long-term trajectory, suggesting that current framework usage is largely maintained for human readability and review, as well as for the rich ecosystem of pre-built integrations (auth, e-commerce, etc.) that have been developed around these frameworks. He notes that for purely static sites, he's already writing more vanilla HTML than he has in a decade.

Both hosts acknowledge that no-code platforms are aware of the threat and are already reacting — Webflow announced a partnership with OpenAI Sites the day before recording, suggesting that established platforms are trying to integrate agentic capabilities rather than be displaced by them. Mike argues that any company serious about long-term growth will have to adapt, and that the transition among working developers has already happened faster than expected — every developer he's personally spoken to in the last four to six months has moved from writing code to reviewing AI-generated code.

The episode closes with Matt introducing the 'coma test' as a way of gauging whether a change is significant enough to matter — and both agree that the AI shift in web development absolutely passes that test, concluding that anyone waking up from a three-year coma and trying to open a code editor would be completely lost in the new landscape.