why is there only a lifetime pricing option for prodigy? You can also consider having a hosted version of Prodigy that people can subscribe with less commitment

I have been looking into using spacy for an entity linking pipeline for my startup, and there's no way to know if it is good or not without trying it on my own data, which means that I have to annotate. Prodigy seems like a good tool to use for annotation, but it only has lifetime licenses. That is a pretty high price tag to pay for something that I just want to try to see if it works for me, especially since I'm running a start up right now and need to be self conscious.

The thing is I'm even happy to pay for a month or something just to try it out, but there is no such option! Obviously there's no way to get a wheel back once someone downloads it, so maybe that's why. But perhaps you guys can have a hosted version that people can subscribe to and try out first.

It must be a common problem that someone needs to try a tool on their own data before committing to it -- in data science you just never know until you try it on your own data, and the barrier to trying is way too high right now.

Hi! I'm not sure I'd call a one-time payment of $390 for an unlimited lifetime license a "high price tag" – at least not on the scale of AI developer tools. Prodigy is downloadable software that you can install on your own machine and run on your own data – to use it, you really want to be programming with it. There's not really a way to do this with a public hosted service or SaaS-only platform. That's why we made it a lifetime licensing model: you may pay slightly more upfront, but you won't have to commit to a subscription payment for a long period of time just so you don't lose access to the software. Instead, you get to own what you paid for.

For commercial trials, we do provide a time-limited VM with Prodigy pre-installed so people can try it out.

You definitely don't need Prodigy to use entity linking in spaCy btw – it'll certainly help, but if you just want to train a model on your own data, you can also annotate it manually, in a spreadsheet, text editor or however you like. It may be less efficient, but it'll certainly work for a proof of concept.