Real-time AI forecasts

Dynamic probability updates

Executive Dashboard

Strategic Foresight at AI Speed & Scale

We replace slow, costly predictive analytics with a scalable AI platform that empowers strategic analysts to continuously forecast market shifts, competitive threats, and emerging opportunities—at the speed of AI

Partnered with

48%

chance

Moderate decline in revenue and customer base

29%

No significant change in revenue and customer base

27%

Moderate growth in revenue and customer base

25%

Significant decline in revenue and customer base

11%

Significant growth in revenue and customer base

1,500 kg to 1,799 kg

53%

1,200 kg to 1,499 kg

26%

1,800 kg or more

15%

1000 kg to 1,200 kg

Less than 1000 kg

H1 2027

40%

H2 2027

29%

2028 or later

19%

H2 2026

10%

H1 2026

Strong but fragmented tightening (divergent rules by region)

52%

Moderate tightening (some regulations on production and usage)

24%

Strong and harmonized tightening (broadly aligned across regions)

16%

Status quo / Minimal change

Swipe to explore

"
« The Forecasting Machine can help conduct horizon scanning at scale and speed which hasn't been possible until now. »
Rafal Kierzenkowski · Head of Strategic Foresight Unit @ OECD

"
« The Forecasting Machine is a great tool to convert an avalanche of news into a collection of time series that describes the world in a far more structured way than the original documents. »
Charles-Albert LeHalle · Professor @ Ecole Polytechnique, Department of Applied Math

Architecture

Smart Forecasting Pipeline

The Forecasting Machine combines event-drivers analysis, real- time multilingual online news gathering, and multi-model collective intelligence to generate richly-documented probability forecasts. They are updated automatically custom schedules, or on demand in response to human feedback.

Market

Policy

Tech

Social

Question Analysis

Analyzes question context to identify key drivers and information needs

News Monitoring

Intelligent querying and gathering of relevant news articles

Event Processing

Extracts and processes key events, storing them in a structured database

Multi-model Prompting

Generate independent forecasts using leading models from Anthropic, OpenAI, and Google.

Outcome A65%

Outcome B25%

Outcome C10%

Confidence Level:

High

Reasoned probabilities

Aggregated probabilities with fully-documented rationales

Benefits

A Closer Look

A simple overview comparing our forecasting method with other common approaches.

Benefit	Simple LLM	Super/Crowd Forecasting	Model-Based Forecasting
Universal Forecast anything, at any time horizon.		Requires experts in every domain and short time horizons	Requires lots of historical data and a separate model for each type forecast
Timely Instant forecasts at AI speed, automatically updated	Requires full prompts and manual updates	Each forecast requires days of research
Documented Easy access to forecast reasoning and sources	Optional	Reasoning and sources, if provided, require synthesis
Scalable Unlimited number of forecasts organized in a rich dashboard	No built-in dashboard	Severely constrained by human cognitive limits	No built-in dashboard
Aware Real-time global news scan informs latest forecasts	No built-in dashboard, limited news scanning	Severely constrained by human cognitive limits	Requires custom news scanning and interpretation
Interactive Forecast updates respond to human input	Requires new prompts	Depends on participant availability	Not inherently interactive
Persistent Retains memory of all past forecasts and events	No persistent memory	Human forecaster turnover results in memory loss
Low Cost Professional quality forecasts for less than a cup of coffee		Expert human forecasting is rare and expensive	Requires costly human model-builder and operator