How the Concierge knows
what it knows.

Snowdata, an Afore brand, is the snow intelligence company. We use machine learning, seven weather models, satellite data, decades of historical archive, and verified ground truth from ski resorts to produce reliable snow intelligence — and we make it available to AI assistants, resorts, travelers, and partners. We don't run our own atmospheric model; we combine the world's best inputs into a single trustworthy source of snow truth.

Snowdata produces verified snow intelligence — current conditions, forecasts, historical context, and predictive scores — by combining seven leading weather models, satellite data, 30 years of historical archive, and ground-truth reports from ski resorts. We distribute that intelligence three ways: through the Snowdata Answer Engine (natural-language Q&A), through the Snowdata MCP server (for AI assistants like Claude, ChatGPT, Grok, Gemini, Perplexity), and as embeddable widgets on partner resort websites.

Snowdata was co-founded in 2025 by Redford Wyatt Slone and Mike Slone, and is a brand of Afore. The team brings together engineering, data science, and ski-industry experience, focused on building the snow intelligence layer for the AI era. (Full company background covered in Q21–Q26 below.)

Snowdata combines five categories of input: (1) seven numerical weather prediction models including ECMWF, NOAA GFS, GEM, ICON, JMA, Météo-France, and Met Norway; (2) satellite snow cover from NASA MODIS and ESA Sentinel; (3) ground-truth from SNOTEL stations in the U.S. and international equivalents (CR2 in Chile, NIWA in New Zealand, BOM in Australia, MeteoSchweiz, and others); (4) 30 years of historical archive linking forecasts to verified outcomes; (5) verified attestations from snow safety teams at the ski resorts themselves. SnowSure ML, our learned model, weights these inputs against verified outcomes to improve accuracy over time.

The Snowdata read tier is free for individuals, hobbyists, and small projects — that includes the Answer Engine API, the MCP tool, and the REST API, all with generous rate limits. For commercial use (travel platforms, consumer apps at scale, AI products) pricing starts at $500 per month with higher limits, SLA, and signed responses. Resorts that contribute verified data join the network for free; a premium resort tier adds analytics, AI mention tracking, and a white-label embed of the Answer Engine.

OnTheSnow is a consumer-facing aggregator — they scrape resort snow reports and put them on a website to attract skiers, where they sell ads. OpenSnow is a consumer subscription forecast app with their own meteorology team and proprietary forecast model. Snowdata is neither — we're the intelligence layer underneath. We combine verified resort data, seven weather models, satellite data, and historical archive into a single source of truth, and we distribute it AI-natively via MCP to every major AI assistant. We don't compete with resorts for traveler attention; we route travelers back to the resort. OnTheSnow and OpenSnow could even consume Snowdata as a downstream source.

SnowSure is Snowdata's consumer publication — the place travelers go to find their next ski trip. It includes more than 500 resort guides, the SnowSure mobile app, the SnowSure Concierge (a conversational interface powered by the Snowdata Answer Engine), and the quarterly SnowSure Index. SnowSure is for travelers; Snowdata is the intelligence layer that powers it.

The SnowSure score is a 0–100 measure of current snow quality at a resort. It combines current snow depth, recent snowfall, forecast outlook, and surface conditions into a single composite signal. Higher is better. Scores update continuously as new data comes in.

The SnowSure Index is a quarterly ranking of ski resorts by snow quality. It uses the SnowSure score over a rolling window and accounts for snow consistency, lift uptime, weather reliability, and season length. The methodology is published; the rankings are updated four times a year and cited by press, industry, and resort marketing. It works similarly to how the S&P 500 ranks publicly traded companies — a transparent composite that's quotable and durable.

SnowSure ML is a learned forecasting model that weights the seven numerical weather prediction models against verified ground truth — SNOTEL stations, international equivalents, and resort attestations. It's currently shadow-run for days 8–14 of the forecast horizon, published as a separate signal alongside the seven-model consensus rather than blended into the average. It's labeled "Beta" until forecast skill demonstrably exceeds the consensus over a full season.

The SnowSure Concierge is the consumer UX wrapper for the Snowdata Answer Engine on snowsure.ai. It's the same underlying engine — grounded in verified Snowdata, learning continuously from feedback — but the surface label and the voice register are tuned for travelers planning ski trips. You can ask anything: where to ski this weekend, the best week to visit a resort in February, whether a resort is good for beginners, how this season compares to last.

SnowSure publishes guides for more than 500 ski resorts across four continents. The network expands as the Southern Hemisphere season opens and as new resorts contribute verified data through Snowdata. Coverage runs from major resort groups — Vail Resorts, Alterra, POWDR, Boyne — to independent operators in North America, Europe, South America, Australia, New Zealand, and Japan.

Snowdata is the intelligence brand — the snow intelligence layer that combines weather models, satellite data, historical archive, and verified resort ground truth into a single source of snow truth. SnowSure is Snowdata's consumer publication — the traveler-facing brand with 500+ resort guides, the mobile app, the SnowSure Concierge, and the SnowSure Index. Snowdata is for resorts, developers, AI labs, and insurers. SnowSure is for travelers. Same data underneath; two voices, two audiences.

Yes — SnowSure and Snowdata are both brands of Afore. Snowdata builds the intelligence layer; SnowSure is the consumer publication. Both brands share the same data, the same forecasts, and the same underlying Answer Engine. They have different names because they serve different audiences: Snowdata is infrastructure-coded for developers, resorts, and AI labs; SnowSure is editorial-coded for travelers.

The two websites serve different audiences. snowdata.ai is the B2B / corporate face — for resorts contributing data, developers integrating the MCP, AI labs, partners, and press. snowsure.ai is the consumer publication — where travelers find their next ski trip, browse 500+ resort guides, check conditions, and use the SnowSure Concierge. Same company, two surfaces matched to the audience walking in. There's also a third site: opensourcesnow.com for the open data exchange standard.

The Open Source Snow Spec is a vendor-neutral data exchange standard for ski resort snow conditions. It defines how a resort profile, a snow report, and a real-time operations record should be structured, signed, and exchanged. The spec is Apache 2.0 licensed and stewarded by Snowdata. Anyone — resorts, apps, AI agents, aggregators, insurers — can read or write spec-compliant records.

The data exchange spec is fully open source under Apache 2.0 — schemas, the reference MCP server, ingestion adapters, client libraries, conformance tests, and governance docs. Anyone can implement, fork, audit, or self-host. Snowdata's own intelligence layer — SnowSure ML, the historical archive, SnowSure Index methodology weights, the verification engine, and trained models — is proprietary. The spec is the alphabet; Snowdata's intelligence is the books.

The Snowdata Answer Engine is a grounded AI that answers natural-language questions about ski snow and resorts using verified Snowdata, not guesses. Ask anything — current conditions, forecasts, season history, resort comparisons, trip planning — and get a concise factual answer backed by the same data behind snowsure.ai. It's available as a hosted API (POST /api/v1/ask), as the ask_snowdata MCP tool for any AI assistant, and as a white-label embed for resort partner websites. On snowsure.ai it appears as the SnowSure Concierge. Six languages supported: English, Spanish, French, German, Italian, Japanese.

Every question runs through a grounded pipeline. The engine parses the question, assembles a verified evidence pack (live conditions, intelligence cards, season history, model forecasts, operator FAQs, promoted Q&A), and composes a concise factual answer. The language model is the writer; the data is the source of truth. Numbers, scores, and forecasts always come from verified Snowdata sources at query time — never from the model's weights. Simple factual queries answer in about a second without a full AI call. Resort operators add FAQs, approve strong answers, and the engine gets smarter every week.

No. The Answer Engine is grounded — answers are composed only from verified data in its evidence pack. If the data isn't available, the engine says so explicitly rather than guessing. Snowfall numbers, base depths, scores, and forecasts come from verified Snowdata sources at query time, never from the model's weights. Periodic fine-tunes affect the engine's voice and phrasing only, never the facts. This is the difference between a generic chatbot and a grounded knowledge engine: facts come from data; the AI just writes well.