plugins and language enabled APIs.

why are language enabled APIs important?

are plugins used for application modernization?

Yes, plugins and language enabled APIs are being used in application modernization efforts but it’s common for new application development, too.

how does Imprompt help developers?

Developers convert their APIs into the OpenPlugin format to enable chatting, testing, monitoring and publishing.
‍
And if the developer is feeling daring, they might just sprinkle on some multimodal extensions like speech and video for more advanced chatting capabilities.

what is function calling?

Most LLMs offer a “function calling” capability. This enables you to ask an LLM to provide the code necessary to to dynamically call an API. The LLM function calling providers focus on three tasks:

1. Collect a corpus of API schemas and calling examples.
2. Finetune a function calling model via the corpus.
3. Build a new inferencing API specifically for function calling.

what are LLM plugins?

Plugins (and specifically, OpenPlugins) are the layer on top of Function Calling routines. Plugins are hosted by plugin servers and invoke the functions. In addition, plugins provide:

1. An abstraction layer over LLM specific function calls. Build once, run anywhere.
2. Standardized documentation for plugin stores and catalogs (authors, vendors, versions, authentication, usage examples, licenses, documentation links, reviews, stars, etc.)
3. Links to the test suites for validation.
4. Multimodal input and output extensions. These are convenience methods to quickly add voice, video, formatting, and generative UI capabilities to the API.
5. Consistent identifiers to track usage and monitor performance (accuracy, latency, …)
6. Extensions to add new capabilities like monetization.

how does OpenPlugin compare to LangChain?

LangChain is a popular developer frameworks for creating LLM and agent based solutions. Both can invoke a function call but OpenPlugin isn’t intended to be a general purpose LLM toolkit. It focuses purely on delivering reliable language enabled plugins, across LLMs, consumable by agents and humans, and discoverable via catalogs.

It’s common for people to build Agents, RAG, etc using LangChain or another framework. We recommend building your language enabled APIs with OpenPlugin and then calling them from your agent app, chat orchestration, or copilot.