Company-based architecture
Each Company keeps its own configuration, prompts, context, tools, and data contracts cleanly separated.
Open-core AI framework
Build AI assistants connected to your business.
IAToolkit is an open-core framework for assistants that query SQL, use internal documents, execute tools, and operate with real business context instead of isolated chatbot prompts.
- Connect operational databases, documents, and APIs.
- Model a Company with its own prompts, rules, schemas, and tools.
- Deploy assistants for real workflows, not only conversations.
Prefer to see something running first? Jump to the demo
Explore the Product Tour
What It Is
Build AI assistants connected to real business context.
IAToolkit helps teams create assistants that reason over data, documents, and tools inside a real Company model, making them useful for actual operations rather than standalone prompts.
Structured intelligence layer
The assistant interprets intent, orchestrates retrieval, and keeps answers aligned with business rules.
Built for trustworthy answers
Ground responses in SQL, documents, and controlled tools instead of asking the model to improvise everything.
How It Works
From data to operations in 3 moves
Connect your sources
Bring in SQL databases, internal documents, APIs, and domain context.
Model your Company
Define prompts, schemas, tools, branding, and business rules around a real operating domain.
Deploy and evolve
Put the assistant in front of users, validate workflows, and iterate with production feedback.
Demo
Explore a sample Company before you decide.
The demo lets you walk through the fictional company Northbridge Consumer Products so you can understand what the assistant sees, how the experience feels, and how the architecture translates into a real product.
- Functional assistant experience
- Business-shaped sample data
- A concrete example of the Company model
Choose Your Path
Two ways to start with IAToolkit
Start with the open-source foundation, then move to the operational layer when the system needs scale and governance.
Build the first serious assistant.
MIT-licensed, self-hosted, and designed for teams who want transparency, control, and a clean implementation path.
- Open-source core
- One Company
- SQL + basic RAG
- Custom Python tools
- Self-managed deployment
Operate assistants in production.
Add the operating model: GUI-driven administration, multi-Company support, governance, and a faster path to scale.
- Admin console for prompts, RAG, SQL, and tools
- Operational configuration and integration packs
- SSO-ready access patterns and governance
- Commercial support and continuous updates
- Available as hosted onboarding or enterprise deployment
Built From Practice
LinkedIn