WITAL — Weather In Text, Any Language, powered by AI

Hyperlocal weather for over 100,000 towns and cities worldwide, delivered as plain-language text in 75+ languages, alongside tabular, icon and chart views for quick decisions.

What WITAL is

WITAL is a free weather companion that turns raw numerical weather prediction (NWP) data into short, human-readable forecasts you can actually use. Instead of a wall of numbers, you get a sentence or two that summarises the day — temperature range, rain timing, wind, and anything noteworthy — written in your language and tuned to your audience. The underlying data comes from the same physical atmosphere models that national meteorological offices run, including ECMWF-AIFS, NCEP-GFS and DWD-ICON, but the rendering layer is built for mobile reading.

Four ways to read the forecast

Every location offers the same data in four complementary views, selectable from a toggle at the top of the forecast screen:

• Text view — an AI-written narrative summary of the day. Short by default, with a control to request more detail. Supports 75+ languages, including Traditional Chinese, Simplified Chinese, Japanese, Korean, Arabic, Hindi, Portuguese, Russian, Spanish, French, German and many more.
• Table view — a compact 3-hourly table covering temperature, relative humidity, wind, cloud cover and precipitation. Past hours are shown dimmed so the present sits in context.
• Icon view — a glanceable 3×3 grid of weather icons across the day at key hours (03, 06, 09, 12, 15, 18, 21, 24). Small emoji hints flag umbrellas, jackets, exercise opportunities or hot/cold at a glance.
• Chart view — a twin-axis time-series: temperature in solid line (°C, left axis) and relative humidity in dashed line (%, right axis), with weather icons plotted along the x-axis. Past hours are shaded so the shape of the day is immediately visible.

Weather models behind WITAL

WITAL aggregates multiple NWP models and, where you enable more than one, computes an ensemble mean for each variable before rendering. Supported models include:

• ECMWF-AIFS — the European Centre for Medium-Range Weather Forecasts' AI Forecasting System. Machine-learning-based, notable for accuracy in the 3–10-day range.
• NCEP-GFS — the US National Weather Service's Global Forecast System. Physics-based, globally consistent, updated four times daily.
• DWD-ICON — the Deutscher Wetterdienst's ICOsahedral Nonhydrostatic model. Strong over Europe at fine resolution.

You can pick a single model or stack several for an ensemble. The icon, chart and event-suggestion views become available when cloud-cover data is present (ECMWF-AIFS, NCEP-GFS, DWD-ICON, or any combination).

Languages and global coverage

WITAL's location database includes over 100,000 towns and cities — down to settlements of a few hundred people in most countries. The text forecast is generated in your chosen language using large language models (OpenAI, Anthropic or Google Gemini, selectable in Settings), grounded in the same underlying numerical data so the physical content stays consistent regardless of language.

Events and context

For selected locations, WITAL surfaces festive, cultural and sporting events happening on the forecast day — handy when travelling or planning outings. A "lifestyle" view suggests actions worth considering given the weather (for example: bring an umbrella, hang clothes out to dry, consider indoor exercise, or heat / cool the home).

Notifications

Once signed in, you can create trigger-based notifications that arrive over push, email or Telegram when a weather condition of your choice is expected at a location of your choice on a schedule of your choice. Useful for commuters, sailors, event planners, gardeners and anyone who'd rather be nudged than check every morning.

Privacy, cost and who we are

WITAL is free to use. The forecast itself is generated server-side and cached, so repeated requests don't re-spend LLM calls. We don't sell your data or track you across other sites. A full privacy policy is available at wital.ai/privacy, and terms at wital.ai/support. WITAL is built and operated by WooSee Limited, a company registered in England and Wales at 71-75 Shelton Street, Covent Garden, London WC2H 9JQ.

Frequently asked questions

Do I need an account?

No. The forecast works anonymously. An account is only required if you want to set up notification triggers or sync preferences across devices.

Which forecast is "best"?

It depends on region and horizon. ECMWF-AIFS is generally strong globally in the 3–10-day range; ICON shines in Europe at short range; GFS gives consistent coverage everywhere. For most users, enabling two or three models and letting WITAL ensemble-average them yields the most balanced forecast.

Why show past hours?

Context. Knowing what the temperature was at 6am makes "18°C at 3pm" more meaningful. Past hours are dimmed so there's no ambiguity about which way time flows.

How often is the forecast updated?

Model runs refresh every 6 hours (00Z, 06Z, 12Z, 18Z UTC). WITAL uses the most recent run's data, which in practice means the forecast you see is at most ~6 hours old.

Is there an iOS or Android app?

An iOS app is available on the App Store. Android users can install the PWA directly from wital.ai on Chrome (Add to Home Screen) for a near-native experience.