Rapid Development Of Intelligent Generative AI APIs

There are instances where merely exposing an API to an LLM does not suffice, and some intelligence needs to be added to the API. Here is a no-code example from an AI productivity suite on how to achieve that…

Introduction

An intelligent API typically refers to an application programming interface (API) that leverages artificial intelligence (AI) technologies to provide advanced functionalities and insights.

These APIs are designed to enable developers to integrate AI capabilities into their own applications, platforms, or services without needing to build the AI components from scratch.

Intelligent APIs can significantly enhance the functionality and value of applications across various industries, including e-commerce, healthcare, finance, customer service, and more. They empower developers to create more sophisticated and intelligent software solutions by leveraging the power of AI without requiring deep expertise in AI algorithms or technologies. Thus accelerating the adoption of AI technologies across various industries.

GALE: Beta Signup

Beyond LLM APIs

Considering the image below, there are instances where directly exposing the LLM via an API does not suffice. These are instances where some kind of model orchestration or intelligence needs to be added to the API.

This logic and intelligence layer can vary in complexity and functionality.

Practical Example

The image blow shows how a flow can be built which serves as the logic for an intelligent API. The flow has input variables defined, which is in turn passed to an LLM node and finally the result forms part of the flows output.

It is evident how different LLMs can be orchestrated and different conditions can also be set.

Source

And finally, below the Python code of the API is executed via a notebook.

import requests
import json

url = "https://gale.kore.ai/api/v1/process/a-1deb775e-0429-5184-ad0b-2cfba9f38a18"

payload = json.dumps({
   "input": {
      "Num1": 1,
      "Num2": 2,
      "Num3": 3,
      "Num4": 2
   }
})
headers = {
   'x-api-key': '$KOREAI_API_KEY',
   'Content-Type': 'application/json'
}

response = requests.request("POST", url, headers=headers, data=payload)

print(response.text)

And the response:

{"startTime":"2024-03-19T20:10:45.646Z",
"endTime":"2024-03-19T20:10:47.818Z",
"elapsedTime":"2.172 seconds",
"output":
{"Result":
" Sure, I'd be happy to calculate that math problem for you. 
59 + 40 + 63 = 162"}}

In Conclusion

The GALE platform can be considered as the Swiss Army knife of the enterprise, capable of automating mid-office and back-office tasks with precision.

At the core of this AI productivity suite you will findworkflow automation, a cornerstone of digital transformation in business processes.

With generative AI at the helm, tedious and time-consuming tasks are seamlessly automated, freeing up valuable human resources to focus on strategic initiatives and creative endeavours.

Workflows can be based on a whole host of modes, external models, fine-tuned models. Or as seen below, an ever growing collection of open-source models…

⭐️ Follow me on LinkedIn for updates on Large Language Models ⭐️

I’m currently the Chief Evangelist @ Kore AI. I explore & write about all things at the intersection of AI & language; ranging from LLMs, Chatbots, Voicebots, Development Frameworks, Data-Centric latent spaces & more.

LinkedIn

Get an email whenever Cobus Greyling publishes.

GALE: Beta Signup