The End of No Code

Bending Spoons buying Airtable for $1.28 billion sounds like a win, and financially it probably is. But a former Airtable engineer is using the deal as a marker for something bigger: the moment no-code platforms stopped being the answer to "how do regular people build software." His argument isn't about Airtable's quality. It's about what changed underneath it.

For two decades, the story was simple. Spreadsheets hit a ceiling on permissions and automation, so teams graduated to tools like Airtable, FileMaker, or SharePoint, which let non-engineers define a data model without touching a database. IT departments, meanwhile, mostly said no to custom software after Salesforce declared "Software is Dead" back in 2000, pushing everyone toward buying SaaS instead of building it. No-code filled the gap between what IT would approve and what teams actually needed, quietly becoming load-bearing infrastructure nobody could rip out.

The pitch now is that large language models with tool use have erased the reason no-code existed. Instead of learning Airtable's column types or wrestling with AWS's alphabet soup of VPCs, IAM, and EBS just to get a tool on the internet, you spin up a Linux VM and let an AI agent build the app directly, in production, from a description of your workflow. The author name-checks a boring, fully open-source stack, SQLite, Go, TypeScript, maybe Vue, and argues the point isn't the stack itself but that lock-in disappears when an agent can port anything to Linux in a few prompts anyway.

He's also selling something: a platform called exe.dev that packages exactly this, cheap VMs, a built-in agent named Shelley, and default-secure sharing, for $20 a month. That commercial angle matters, because it means the argument doubles as a product pitch dressed up as a manifesto. Still, the underlying claim is worth sitting with: if an LLM can one-shot a Slack bot, wire up cron jobs, and debug its own performance with a profiler, the value no-code platforms provided, letting non-engineers specify structure without code, gets absorbed into a conversation with an agent instead of a UI with dropdowns.

Whether that holds up outside a blog post is another matter. Vibe-coded internal tools built by an AI on a $20 VM will accumulate the same mess spreadsheets always did, just faster and less visibly. But the direction is clear enough that a $1.28 billion acquisition of the category's most beloved product reads, to at least one insider, like a going-away party rather than a coronation.