About CropIntel
CropIntel is built and maintained by Alastair Campbell through Hurricane Works. It started as a research project in early 2026 and now runs daily in production, fully public, every call, the per-region breakdown, and the forward-prediction log are open, no login.
The outputs are public by design. What's not public is the methodology, the curated UK arable practitioner corpus, the lexicon, the model internals. That's the defensible IP. You can audit every call we've made; you can't rebuild the engine from the public site.
CropIntel serves grain merchants, crop insurers, and commodity desks (see the value to each). For an acquisition conversation about the engine, corpus, and track record: alastair@hurricane.works.
About the author
Alastair Campbell built CropIntel after fifteen years building data products in adjacent industries. He runs Hurricane Works, a small studio that has shipped consumer-facing data products covering vehicle pricing, used-car valuations, and B2B market intelligence. Prior product work includes Carsnip and ongoing operator roles in automotive data.
The CropIntel project is informed by, but not derived from, that prior work. The pattern is consistent: a thinly-instrumented market with fast-moving sentiment and slow-moving official data, where a daily, region-by-region directional call earns its keep relative to monthly outlook commentary.
Day-to-day operations: data ingestion, model maintenance, methodology documentation, and any buyer conversations are handled directly by Alastair. There is no team dilution between buyer and builder. The system itself runs autonomously on a documented daily pipeline, so what transfers is a working production system with a runbook, not a person's calendar.
External: Hurricane Works · LinkedIn
Why this exists
UK wheat is a £2bn+ crop with a price formation that's surprisingly late-binding. By the time DEFRA publishes harvest figures or AHDB releases its monthly condition reports, the trade has already repriced. Compound weather stress, satellite canopy data, and real-time farmer reports give a directional view weeks earlier, at high enough accuracy (62.3% on a walk-forward backtest) to be useful for actual decisions: forward buying, hedging, insurance pricing, and parametric reserving.
The system is intentionally narrow: UK wheat, daily cadence, directional calls only. It is not a multi-commodity global agronomic platform and has no plans to become one.
Contact
Acquisition enquiries, press, or research questions: alastair@hurricane.works
The model, methodology, curated corpus, public track record, codebase, and brand are available to acquire by a strategic buyer. See the opportunity for the commercial model.
What we do not provide
- Regulated investment advice. CropIntel is a data product that publishes directional forecasts. Trading involves risk of loss.
- Multi-commodity coverage. UK wheat is the focused product. Barley, OSR, and feed grains are tracked but not separately published.
- Guarantees of accuracy. The 62.3% historical hit rate is statistically validated but no system predicts the future perfectly.
- Real-time alerts (yet). The product cadence is daily. Alerting tooling for paying customers is on the roadmap.
Related: Methodology · Track Record · The opportunity