Welcome to aiAgent

The aiAgent library offers a seamless interface to interact with Natural Language Processing (NLP) and Semantic Network (SN) AI systems. It empowers users to easily upload, manage, and contextually query content across various AI systems. It also makes it easy to integrate advanced AI features into an application,

The aiAgent library simplifies querying and managing content stored in distinct, isolated storage areas. Users can create collections, such as “general docs,” “email,” or “calendar,” and query them individually or in combination.

Each user can access their own storage set as well as the storage of the user parent hierarchy. The user parent hierarchy structure allows for different levels of shared knowledge to be exposed to users of the given installation.

Example Usage Scenarios

Enterprise company internal knowledge base

A large enterprise organization uses aiAgent to maintain a comprehensive knowledge base across the entire enterprise, top to bottom, that can be queried by all members.

There can be an aiAgent user assigned as the top enterprise-level organization that has content related to company standards, human resources guides, and more. Subgroups within the enterprise can also create their own knowledge repositories that are specific to the members of their subgroup. Each user has their own knowledge base only they can see.

This structure enables users to access and query AI-generated data tailored to their specific needs, encompassing their individual role, subgroup, and the larger enterprise. A user query can be answered from knowedge stored at any of the levels they are related to through the user parent hierarchy.

Production system log analysis

A large, production deployment of a public facing system needs advanced log analysis in order to recognize user trends, error states and other activities in order to take customized action on. Doing log analysis by having each log statement pushed into a “log” storage so that AI queries against log statements makes it easy for queries about the state of the system to be answered.

Using a hugging face model and pretrained dataset to start with allows for log data to be kept local and be processed in volumes and ways that would be costly through systems like chatGPT.

Benefits of Storage Solution

Keeping storage data isolated provides a collection of benefits such as:

  • easy to control security around access to specific storage.

  • easy to remove a user by removing their storage rather than trying to remove rows from a larger db

  • easy to move users to different servers with various sharding methods including a basic usage of python’s hash_ring library.

Building on the Work of Others

This project wouldn’t be possible without the ground breaking work done by other teams and projects including:

  1. OpenAI ChatGPT - Version 3 of their system was really caused the world to take notice and is the core reason many of the capabilities of aiAgent are possible.

  2. LangChain Python Framework/Library - This project provides a critical set of features that build on top of ChatGPT and other AI systems thanks to its robust integration with the hugging face models and libraries through its use of pytorch.

  3. Pytorch - This project provides key neural network libraries that allow data to be transformed into tensors and used to train the neural net and validate it. There are now many libraries that work together to make building AI systems with Pytorch powerful yet approachable.

  4. HuggingFace - A community of big names bringing models and datasets to be used for free in an open way to make AI more accessible.

Indices and tables