← @bgSponge

Extensible Software in the age of LLMs | Jeremy Morrell

Summary

Jeremy Morrell argues LLMs enable 'Software for One' by allowing users to extend stable software cores through natural language prompting, bypassing the complexity of traditional feature bloat.

  1. Feature bloat ruins usability.

    Adding features for small user segments increases complexity for everyone else, potentially making the product worse for the majority.

    user interfaces can only become so complex before they become unusable

  2. LLMs enable personal applications.

    Users can now speak code into existence to create custom workflows that side-step enterprise software complexity.

    your users have suddenly acquired the ability to speak code into existence

  3. Object-capability models improve security.

    Removing ambient I/O ensures code only acts via passed references, which is easier to reason about than proxy logic.

    the code can only take actions via the references it has been passed

  4. WebAssembly provides a secure sandbox.

    WASM is an attractive candidate because it lacks built-in modules for I/O like HTTP requests or environment variables.

    there are no built-in modules for making an HTTP request

  5. Extension authorship remains unequal.

    Even with LLM acceleration, Morrell expects a small percentage of users to author most extensions in any ecosystem.

    A small percentage will author most of the extensions

Read the original ↗Save to your Sponge

Summarized with Sponge. Quotes are the author’s own words.

More from @bg

All readings →