Overview
Giotto is a deployable intelligent system with agentic capabilities. It is designed to run as a complete AI system, combining language-model intelligence, tool use, orchestration, and observability in a single platform.
A Giotto deployment can run even on a single H100 GPU, making it possible to operate a capable intelligent system without requiring a large multi-node setup from the start.
How users interact with Giotto
The two main ways to interact with Giotto are:
Chat interface
The simplest way to use Giotto is through the Chat interface. This provides a user-friendly conversational entry point where users can submit requests and observe the system’s agentic behavior. See the article How to Use the Chat Interface for more details.
Python client
Giotto can also be accessed programmatically through the Python client. The client follows an OpenAI-style API, making it familiar for developers who already work with OpenAI-compatible workflows. See the article Giotto Python API Library for more details.
This means Giotto can be used both:
- interactively by end users through chat
- programmatically by developers and applications through Python
Platform architecture
Giotto is more than a single model endpoint. It includes the infrastructure needed to deploy, run, monitor, and operate intelligent workflows.
Its architecture includes:
- computation and serving components
- orchestration components
- logging and monitoring systems
- tracing and metrics tools
- storage and access-control layers
This gives users not only an AI runtime, but also the operational tools needed to understand and manage it.
In simple terms
A good way to think about Giotto is:
Giotto is a deployable AI system that can reason, use tools, and be operated like a real production platform.
It combines:
- Intelligence
- Deployment
- Orchestration
- Observability
into one system that can be used through chat or through code.
Comments
0 comments
Article is closed for comments.