TimeCopilot/timecopilot
TimeCopilot: the GenAI Forecasting Agent. Built on LLMs and Time Series Foundation Models, it lets you forecast, cross-validate, and detect anomalies using multiple foundation models through a single API. From finance and energy to web analytics, TimeCopilot turns natural-language queries into production-ready forecasts.
What it solves
TimeCopilot simplifies complex time series forecasting by automating the selection of the best forecasting model and providing natural-language explanations for the results. It removes the need for users to manually write scripts, build pipelines, or create dashboards to perform professional-grade time series analysis.
How it works
It acts as a generative agent that combines Large Language Models (LLMs) with over 30 time-series foundation models (such as Amazon Chronos, Google TimesFM, and Salesforce Moirai). The LLM interprets statistical features of the data, guides the selection of the most accurate model based on data characteristics, and translates technical decisions and forecasts into plain English.
Who it’s for
It is designed for users who need accurate time series forecasts but may lack the deep statistical expertise to manually select and tune models, as well as developers looking to integrate automated forecasting into their applications.
Highlights
- Unified Forecasting Layer: Integrates 30+ foundation models with LLM reasoning for automated selection.
- Natural-Language Interface: Allows users to perform forecasts and ask specific questions about future trends and patterns using plain English.
- One-Line Execution: Supports end-to-end forecasting from a web URL using a single command via
uvx. - Comprehensive Analysis: Provides not just predictions, but also statistical feature analysis, model comparison results, and reasoning for the chosen model.