I’m one of the developers at Roadmap Technologies. We recently made the Geneva Forecasting API available to developers, with a free tier that includes 1,000 forecasts per month.
The API is built around the Geneva forecasting engine, which has been used in Oracle planning and forecasting environments for years.
The goal is to make time-series forecasting easier to integrate into applications and workflows without teams having to build and operate the forecasting infrastructure themselves.
It currently includes:
- automatic model selection - forecast playground - prediction intervals - support for single and batch forecasting - docs, examples, and developer portal access - MCP support for agent workflows
On our current setup, single forecasts typically return in under 200 ms, and batch forecasting can process hundreds of time series per second.
Access is currently approval-based while we roll things out.
Happy to answer questions about the models, latency, batching, infrastructure, and where this does or does not make sense compared with open-source tools.