AI-ready
The Timeseries Refinery includes a built-in MCP (Model Context Protocol) server. Connect any MCP-compatible AI assistant directly to your time series catalog — and let it explore, compute and analyse your data in natural language, with full traceability and governed access.
Search for series by name, metadata or custom queries. The AI navigates your full time series catalog and finds exactly what you need — without knowing the exact series names in advance.
Retrieve time series with flexible time windows, automatic resampling and statistics. Ask business questions like "compare forecast vs. actual production over the past week" and get an immediate analysis.
Automatically detect change points, characterise regimes (ascending, oscillating, chaotic, mean-reverting…) and flag spikes. The AI explains what it finds in plain language.
The AI can write, test and register new formulas in the Refinery's formula engine — not as one-off Python scripts, but as versioned, auditable, reusable calculations that any team member can inspect.
Instruct your AI assistant to build or update operational dashboards and figures directly in the Refinery — from data to visualisation in a single conversation.
The MCP server enforces the Refinery's role-based access rights. Your AI assistant only sees and modifies what you are authorised to access — no data leaks, no bypassing governance.
The Timeseries Refinery is an open-source platform for storing, computing and visualising versioned time series data — built for data-driven teams in energy, trading and industry. Its bitemporal storage keeps every version of every series: any series can be queried as it was known at any past date (as-of queries) — and computed series follow the same law: formulas evaluate against the versions known at that date. It provides a traceable formula engine, real-time dashboarding, an Excel client, and full Python and REST APIs. Learn more