PromptHub · coming soon

AI prompts as infrastructure, not strings in code

Write prompt templates with Handlebars syntax and typed variables. Planned for launch: type-safe SDK functions for Go and TypeScript, and prompt updates without redeploying your app.

Handlebars templatingTyped variablesPrompt generatorSDK generation (planned)Version history (planned)Coming soon
SDK generation · planned

Type-safe functions, not string building

At launch, PromptHub will generate SDK functions for Go and TypeScript, with Python next. Each function takes typed arguments that match your template variables, so your AI prompts get compile-time safety.

  • Go structs and TypeScript interfaces first, Python dataclasses next
  • Compile-time type checking catches errors before runtime
  • Import and call, with no template parsing in your code
See the roadmap
API & runtime · planned

Update prompts without deploying

Use the generated SDK for compile-time safety, or fetch prompts over a REST API for runtime flexibility. Change a prompt in the dashboard and every API call will return the new version right away.

  • REST API with API key authentication
  • Update prompts without code changes or deploys
  • A/B test prompt variations without feature flags
Join the waitlist

From prompt to production code

Templates, typed variables and the prompt generator work today. Everything else on this list is planned for launch.

Handlebars templating

Write prompts with real templating syntax: conditionals, loops and context switching, not just variable interpolation.

Handlebars templating

Write prompts with real templating syntax: conditionals, loops and context switching, not just variable interpolation.

Use {{#if}}, {{#each}}, {{#with}}, and {{#unless}} blocks in your prompts. PromptHub parses the template, detects all variables, and understands the structure. Your prompts become proper templates, not fragile string concatenation.

Typed variables

Detected variables get a type: text, number, true/false, list, object or options. At launch, types will drive the SDK and catch errors before runtime.

Typed variables

Detected variables get a type: text, number, true/false, list, object or options. At launch, types will drive the SDK and catch errors before runtime.

When you write {{name}} in a prompt, PromptHub detects it and lets you assign a type. At launch, these types will flow into the generated SDK functions: a string stays a string, a boolean becomes bool, and arrays become slices. Type mismatches will be caught at compile time, not when the AI returns garbage.

Type-safe SDK generation · planned

You'll download generated functions for Go and TypeScript, with Python next. Each will take typed arguments and return the formatted prompt text.

Type-safe SDK generation · planned

You'll download generated functions for Go and TypeScript, with Python next. Each will take typed arguments and return the formatted prompt text.

Define your prompt template, assign variable types, and PromptHub will generate a function in your language. For Go: a struct with typed fields and a function that returns the prompt string. For TypeScript: an interface and a function. No template parsing at runtime, no string building, no type guessing.

Version history · planned

Every edit will be versioned, so you can compare any two versions side by side and roll back to an earlier one with one click.

Team collaboration · planned

A shared prompt workspace for your team, where everyone sees the same prompts, the same versions and the same variable types.

API-first · planned

Fetch prompts through a REST API at runtime, update them without redeploying and A/B test variations without code changes.

API-first · planned

Fetch prompts through a REST API at runtime, update them without redeploying and A/B test variations without code changes.

Use the generated SDK for compile-time safety, or fetch prompts via API for runtime flexibility. Update a prompt in the dashboard and every API call returns the new version immediately. Run experiments by serving different prompt versions to different users, with no deploy needed.

The problem

Why this needs to exist

AI prompts today are strings buried in code. No versioning, no types, no collaboration. Every team reinvents the same bad patterns.

Prompts in code are fragile

String concatenation, f-strings, template literals. One wrong variable and the AI gets garbage input. No compiler catches it. No tests cover it.

No single source of truth

Prompts scattered across repos, Notion docs, Slack messages. Product wants to tweak copy, but they need a developer to deploy it.

No iteration workflow

Change a prompt and hope for the best. No version history, no diff, no rollback. Was the last version better? Nobody knows.

Be the first to try PromptHub

PromptHub is in active development. Join the waitlist and we'll let you know when it's ready.

  • Coming soon
  • Go & TypeScript first
  • Handlebars
  • Free tier at launch
// tick. tick. tick.
0 ticks since founding
Coming soon

Join the PromptHub waitlist

PromptHub is in active development. Leave your email and we'll invite you as soon as it opens, with a free tier and no credit card.

One email when access opens. No newsletter, no spam.