Faim FAQs

Question 1

What is Faim and its primary purpose?

Accepted Answer

Faim is an MCP (Model Context Protocol) server designed to integrate the powerful FAIM time series forecasting SDK with any MCP-compatible AI assistant. Its primary purpose is to enable AI-powered time series forecasting capabilities directly within your AI assistant workflows.

Question 2

What input data formats does Faim support for time series?

Accepted Answer

Faim supports flexible input formats for your time series data. You can provide data as 1D arrays for single univariate series or as 3D arrays in a batch/sequence/feature format, accommodating various data structures.

Question 3

What kind of forecasting capabilities does Faim offer?

Accepted Answer

Faim offers comprehensive forecasting capabilities including point forecasts (single value predictions), quantile forecasts (providing confidence intervals with custom levels for detailed risk assessment), and sample forecasts (generating distribution samples for a broader understanding of potential outcomes).

Question 4

Which time series models does Faim utilize?

Accepted Answer

Faim integrates with the FAIM API to leverage advanced foundation time-series models for zero-shot forecasting. Currently, it exposes two powerful models: Chronos2 and TiRex, known for their robust prediction capabilities.

Question 5

How does Faim integrate with AI assistants or other applications?

Accepted Answer

Faim implements the open Model Context Protocol (MCP), exposing `list_models` and `forecast` tools. This makes it compatible with any LLM, AI framework, or application that implements an MCP client, allowing seamless connection for AI-driven forecasting.

Faim

Faim

Key Features

Use Cases

Key Features

Use Cases