Why I'm not trying to compete with SkySight

SkySight is the best soaring forecast product on the market and I am a paying user. I get asked, almost weekly, why I am building a soaring weather API when SkySight already exists - and the honest answer takes more than a tweet. So this is the longer version.

SkySight is a pilot-facing product. You log in, you see beautiful forecast maps, you scroll through soundings, you plan flights, you go fly. Matthew Scutter and the team have spent years tuning the model, refining the UI, and building one of the most loyal user bases in the sport - cross-country and competition pilots in particular swear by it. The product works because every single design decision is optimised for one job: helping a pilot decide where and when to fly. If that is your job too, SkySight is what you want. Convek is not trying to do that job. We do not have a pilot UI. We do not produce forecast maps you can stare at over breakfast. If you came here looking for a SkySight alternative for your weekend flying, this is the wrong product.

The job Convek is trying to do is different. There is a category of people who do not just want to look at soaring forecasts - they want to build something with them. An iOS app for paragliders that automatically scores tomorrow's site list. A vario that suggests a different course based on the predicted convergence line ahead. A club website that shows a thermal forecast widget for the local site. An XC contest tool that surfaces the day's best regions. Every one of those needs the data as JSON, not as a rendered map. None of those people can be customers of SkySight, because SkySight is not built to be queried by software - it is built to be looked at by humans. That is not a flaw, it is the design.

So we are not competing on the same axis. SkySight is the best soaring-forecast UI; we are trying to be the best soaring-data API. The data underneath is in the same lineage (regional WRF runs on global initial conditions, with RASP-style derived fields), but the product shape is fundamentally different. A pilot will use SkySight and barely think about us; a developer will use Convek and barely think about SkySight. The rare middle case - someone building a pilot UI from scratch - might use both, with SkySight as the reference for what good looks like and Convek as the data layer.

On the meteorology, I want to be honest. SkySight has years of operational tuning, deep adoption among top competition pilots, and a feedback loop from a serious user base that is impossible to fake. Their model is excellent. Convek is younger and the validation pipeline is still being built - I have written about that openly in the validation post. I am not claiming the forecast is better. I am claiming it is shaped like an API instead of a UI, which is the relevant axis for the developer audience.

Where the comparison does land cleanly is access. SkySight pricing is per-pilot subscription, designed for individuals. There is no developer tier, no API, no way to integrate the forecast into a downstream product. Convek pricing starts at a free tier (25 queries/day, production-allowed), with paid commercial tiers from £39/month, and the redistribution rights are explicit on the paid plans. If your goal is to build something on top of the data rather than to look at the data yourself, that is the practical difference. The side-by-side comparison page lays out the rest.

If you are a pilot reading this and you do not already have a SkySight subscription, get one - it is genuinely one of the best-value subscriptions in the sport. If you are a developer who has been told by a pilot friend that SkySight is the answer, this post exists to explain why it might not be the answer for your specific use case. Both products are real, both have audiences, and the two audiences barely overlap.