Is Glyph worth it if I'm not using AI?

These are language features, not AI features

Strip the word "agent" out and read the list: no any, exhaustive match, errors as values, types that are checked at runtime, one declaration form per name, a canonical diff-stable format. Anyone who's written Rust, Elm, or OCaml recognizes these as the things that make a codebase pleasant to maintain — they were good ideas long before AI wrote code. Glyph just brings them to a TypeScript-shaped language you can adopt one file at a time.

What you get, solo

• Fewer 3am pages. Boundary data is validated by construction (User.parse), and a missing case is a compile error — not an undefined that surfaces in production.
• Faster review. A one-line change is a one-line diff, so a reviewer reads the change, not the reformatting churn around it.
• You can find anything. grep "fn parseUser" lands on the definition — no overloads, barrels, or decorators scattering it.
• Easier onboarding. A small core with one way to do each thing means less tribal knowledge for the next person to learn.

So why lead with agents?

Because the payoff is bigger there. A human eventually learns to run the linter and handle the case; an agent reaches for any, forgets the arm, and merges it green — and it can't be trusted to keep the opt-in tooling on. The value of making the strictness non-optional compounds when a machine is writing the code. But the floor — the wins above — is there whether you use an agent or not.

Think of it as a stricter TypeScript that happens to be great for agents — not an agent tool you can only justify if you use one.