About
The Forecasting Company builds foundation models and enterprise-grade forecasting systems that predict arbitrary time series for planning and operational decisions. It sells to enterprise teams, differentiating itself through in-house temporal foundation models, large training datasets, and plug-and-play access through the Retrocast browser platform and API.
Market
The company competes in enterprise AI forecasting, time-series foundation models, and operational planning, with applications spanning demand planning, inventory, manufacturing, energy, and supply-chain optimization. It positions itself as a plug-and-play alternative to traditional forecasting systems that require extensive data-science teams, combining broad cross-industry training data with user variables or curated data streams. Its browser platform, REST API, rapid inference, backtesting, and long prediction horizons differentiate it from both specialized model/API providers and larger enterprise planning suites.
The Forecasting Company targets enterprise organizations with forecasting and planning needs, particularly operators in retail, supply chain, manufacturing, energy, and pharmaceutical businesses. Its likely buyers are operations, supply-chain, demand-planning, and data-science leaders seeking production forecasts without building and maintaining large specialist teams.
At a Glance
Problem
Businesses make operational decisions against uncertain, volatile demand, but existing time-series forecasting tools can produce unreliable predictions—especially during spikes such as Black Friday. Building and maintaining a useful forecasting system typically requires expensive teams of data scientists and domain experts, while backlogged data-science teams can leave operators waiting months to put a use case into production. The economic cost is being caught between overstocking, shortages, waste, excess maintenance inventory, and missed sales.
The clearest use case is demand and inventory planning: retailers and manufacturers can forecast product or service demand, keep shelves and production lines supplied without carrying unnecessary stock, and align purchasing, production, pricing, and fulfillment with expected demand. The company also targets adjacent planning problems in energy, transportation, pharmaceutical supply chains, and manufacturing.
Product / Service
The Forecasting Company provides an enterprise-grade, plug-and-play forecasting system built on foundation models for time-series data. Its model is pretrained on a large, cross-industry dataset of time-stamped data, so customers do not need to train a bespoke model from scratch. Users can upload time-series data through a browser-based platform, run one-click backtests, use their own input variables or curated data streams, and generate forecasts ranging from the next quarter to two years ahead.
The delivery model combines a browser platform with an API for integration into existing workflows. The intended benefit is faster deployment of accurate forecasts with less dependence on specialist modeling teams; the company’s site illustrates a sample backtest in which its model achieved a 14.4% WAPE versus a 19.2% baseline, or 25% better performance on that example. Forecasts are designed to reduce inventory and operating costs, prevent shortages and overproduction, and improve sales and supply-chain decisions.
Market
The company operates in AI-powered time-series forecasting, demand planning, supply-chain planning, and enterprise productivity software. Its competitive set includes established planning platforms such as SAP, Kinaxis, Oracle, o9 Solutions, and Blue Yonder, as well as focused forecasting platforms such as Nixtla’s enterprise TimeGPT. The Forecasting Company’s differentiation is its attempt to offer a flexible, cross-industry foundation model that is easier to deploy than conventional, use-case-specific forecasting systems.
The company was founded in 2024, joined Y Combinator’s Summer 2024 batch, and was listed as active with a seven-person team. It publicly launched its product by February 2026 and offers a sandbox, browser platform, and API access application. Public third-party data indicates early commercial traction rather than a purely pre-revenue project: Latka estimated $330,000 in 2025 revenue, while PitchBook records $500,000 raised and labels the company early-stage and revenue-generating; however, no current revenue figure or named customer list is publicly disclosed in the available sources.
Founders & Leadership
Funding History
Y Combinator
Recent News
The Forecasting Company announced its participation in VivaTech 2026 in Paris, where CEO and co-founder Geoffrey Négiar was scheduled to present Retrocast and the company’s forecasting foundation model.
Alongside its VivaTech announcement, The Forecasting Company said it had released an early version of its t₀ forecasting foundation model on Hugging Face.
The Forecasting Company published Part 2 of its “Future of Forecasting” series, discussing why conventional tabular representations can obscure important structural differences in time-series data.
Part 1 of The Future of Forecasting series explains how a forecasting foundation model embedded in an agentic harness could perform the work of human forecasters across millions of time series.
Shoptalk’s sponsor directory listed The Forecasting Company among its partners and linked to the company’s website. The available evidence does not describe a specific integration or commercial partnership.
The Forecasting Company highlighted CEO Geoffrey Négiar’s appearance on Micode’s Underscore_, where he discussed forecasting, foundation models, and the company’s work.
Active Roles
1Business Model
The company appears to use a gated B2B software/API model, offering enterprise customers access to its Retrocast browser platform and unified forecasting API. Users apply for API access or log in; public pricing and specific subscription or usage terms are not disclosed in the available evidence.