NOTE: Once again, Khoros killed my posting mojo. I spent an hour creating a great post only to lose it; some sort of security issue. I am so done with this community platform. I retyped it in my Substack. [sigh]
As you may know, Google released beta access to PaLM LLM APIs this week, and I’ve been experimenting with it. I can’t say I fully understand it, but I’ve successfully built stuff with text embeddings, chat conversations, and text completions. Most of my tests have been in the Google Cloud Platform, but this one was created in Airtable.
Airtable Example
Given a list of target websites in this table (Target Sites), I created a script that extracts a summary of each and places it into the Synopsis column. Then it accesses the data from a specific financial website to extract the number of employees and the company’s market cap based on its stock symbol.
In a sense, this demonstration uses AGI to scrape the web, a term I detest but a concept I love. Is it possible that the days of scraping the HTML topography to extract data is about to end?
