I’m attempting to make a simple digest of email newsletters that I get. I want to send it through OpenAI to summarize each newsletter, and then store and send in a digest.
I’ve built the scenario out, but the problem is that the body of most email newsletters includes a ton of URLs. They’re often super long, and all of the characters end up maxing out GPT-3 token max and erroring.
Is there any way I could add a formatting step to just pull out the plain text with all links etc removed?