Vorschau
Strategic foresight · multi-agent AI

Futures you can defend, not just describe.

Multi-agent strategic foresight for small and medium-sized enterprises.

Vorschau operationalises a 12-stage strategic-foresight pipeline through cooperating LLM agents. It is built for SMEs that need rigorous, evidence-backed foresight without a dedicated futures team.

EU-hosted modelsOpen calibration dataEarly access
Specialists
12
AI roles working as one team
Live sources
10
news, patents, research, filings
Checkpoints
5
the run must pass before advancing
Region
EU
data stays in Europe

What rigour looks like

  • Every signal confirmed by at least two independent sources
  • Genuinely different futures, not four flavours of today
  • Every forecast stored with the date it will be checked against reality
  • A traceable line from source to signal to scenario to strategy

See a real session

Browse three finished foresight runs

Full-depth pipelines on three German Mittelstand questions — mechanical engineering, specialty chemicals, medical devices. Real signals, real scenarios, real strategies. No sign-up.

Public demo account

Sign in and see how it reads

Three sectors, three finished pipelines. Every signal, scenario, strategy and signpost is inspectable down to its source.

Mechanical engineeringSpecialty chemicalsMedical devices

Access is granted on request.

Who it's for

Two audiences, one platform

Read the version of the story that matches what you came here for. Both link to what you need next.

For decision-makers

You run an SME and the world keeps shifting under your feet. You want a structured way to stress-test your strategy against several plausible futures. Not a prediction, a preparedness drill. Sessions take 15 to 30 minutes and produce a one-page brief you can take to your board.

For researchers

You care how multi-agent LLM foresight is actually done. The platform operationalises CLA, Futures Triangle, morphological analysis, RDM and calibration inside a 12-stage pipeline, with hard structural checkpoints, full provenance and a public calibration log.

What it does

Six layers of disciplined foresight

Each capability is a structural commitment, not a prompt that hopes for the best. Every claim is tied back to evidence and, where it's a forecast, to a date it will be checked.

SCAN

Scan the horizon

Ten live sources: news, research, patents, filings, forecasts. Every signal is cross-checked across at least two independent sources before it earns a place in the brief.

INTERPRET

Interpret with depth

The system deliberately reads signals from four angles: the headline, the system behind it, the worldview that takes it for granted, the story underneath. Not just the first obvious take.

PROSPECT

Generate alternative futures

Several futures, not one forecast. Diversity is enforced structurally, so you never get four flavours of today. Different enough to disagree with each other.

CHALLENGE

Stress-test strategies

Every strategy is run through every scenario. You see which ones hold up, which are bets, and which only work if you are lucky. No consensus-future wishful thinking.

LEARN

Calibrate and learn

Every forecast is stored with the date it will be checked. Six months later, the system grades its own accuracy and tells you which agents to trust more, or less.

GOVERN

Keep humans in the loop

You approve at every act boundary. Every claim traces back to its source. Built for SMEs without a dedicated foresight team. You bring the judgment; the system brings the scaffolding.

Ready to try a foresight session?

Sign up in under a minute. A short onboarding captures your context; your first run walks you through framing, evidence and stress-tested strategies in about 20 minutes.