Bridging the Last Mile of Time Series Forecasting with LLM Agents
2026-06-01 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors point out that in real-world forecasting, initial model predictions often need adjusting based on business-specific information like holidays or marketing plans. They call this adjustment step 'last-mile forecasting' and introduce a system using large language models (LLMs) to help revise forecasts by gathering extra context and making controlled updates. Their framework organizes information clearly, uses tools to find relevant data, and keeps track of changes to ensure forecasts are reliable and understandable. They demonstrate that this approach helps turn raw predictions into practical, decision-ready forecasts.
time series forecastingfoundation modelslast-mile forecastinglarge language modelsforecast revisioncontextual evidence retrievalstructural safety constraintslong-horizon forecastingmap-reduce decompositionmemory bank
Authors
Yuhua Liao, Zetian Wang, Qiangqiang Nie, Zhenhua Zhang
Abstract
Time series forecasting has advanced rapidly, especially with the emergence of foundation models that show strong zero-shot performance on numerical extrapolation. However, in real-world forecasting settings, a statistically plausible baseline is rarely the final forecast used in practice. Before a forecast becomes decision-ready, it often needs to be revised using weakly structured business context such as holiday effects, campaign plans, external events, historical analogs, and expert feedback. This practical stage remains underexplored in the forecasting literature. In this paper, we formulate this stage as the \textbf{last-mile forecasting} problem and present an LLM-agent framework that sits on top of a forecasting backbone. Our system maintains a unified forecast workspace, invokes tools to retrieve contextual evidence, and converts reasoning trajectories into explicit forecast revision actions under structural safety constraints. It also supports long-horizon forecasting through map-reduce-style decomposition and post-hoc reflection through a memory bank. The resulting system is designed to be controllable and auditable. Through real-world case studies, we show how LLM agents can bridge the gap between statistical prediction and business-ready forecasting.