{% extends "base.html" %} {% block title %}IAToolkit - AI Framework{% endblock %} {% block styles %} {# Link website stylesheet and Bootstrap icons #} {% endblock %} {% block content %}
IAToolkit
EN | ES

Build real-world AI assistants powered by your own data

IAToolkit is an open-source framework for building AI assistants that operate inside your environment and use your company’s data securely and reliably.

  • Native integration with your corporate data sources.
  • Multiple Companies defined through configuration, each with its own domain.
  • An Intelligence Layer that orchestrates prompts, tools, and RAG coherently.

Try the Platform

Explore a realistic demo configured for a sample trading company. Ask questions over real SQL data and internal documents, and see how a real-world assistant behaves when it is backed by IAToolkit.

100% Open Source

A fully open framework you can audit, extend, and deploy on your own infrastructure – no black boxes, no proprietary lock-in.

View on GitHub {% if iatoolkit_version %} {{ iatoolkit_version }} {% endif %}

Multi-Tenant Architecture

Serve multiple business units or clients from a single installation. Each Company is isolated through configuration, tools, prompts, and context.

Connect Your Data

Connect SQL databases, internal documents, and APIs so your assistant answers with real corporate information instead of hallucinations.

Intelligence Layer

A structured layer that interprets intent, reads each Company’s configuration, orchestrates queries, prompts, tools, and RAG, and keeps responses aligned with business rules.

Multi-LLM

Use different providers (OpenAI, Gemini, local models) without changing your code. Choose the best model per Company or per workflow.

100% Customizable

Adapt branding, tools, prompts, context, and workflows so each assistant reflects your identity and supports your internal business processes.

Foundation / Core Architecture Explained

A clear explanation of how IAToolkit is designed as a framework for assistants that operate in real-world scenarios, not just demos or proof-of-concepts.

Read

Mini-Project / Build Your Internal ChatGPT

A 3-month plan to build your own “internal ChatGPT”: a corporate assistant capable of answering with real data, internal documents, and predefined prompts.

View plan
Photo of Fernando Libedinsky
About the author
LinkedIn

I am Fernando Libedinsky, software engineer and creator of IAToolkit.
After a long career building software products, I am still driven by the same curiosity that led me to write my first lines of code: to learn, explore, and build new things.
IAToolkit is a continuation of that drive: a platform designed to rapidly connect companies with AI in a practical, production-ready way.

{% endblock %}