Hello,
Am I right to conclude that I can’t use the ChatGPT API to receive live data?
ChatGPT on it’s own (using model 4o), can actually get information from the web, but the API does not seem to be able.
I have two questions:
Am I right about this?
Is there an alternative or a way on make.com that will allow their API to gain live data from the internet?
Yes
Yes
You can try searching for “ai scrape website” to see what others have tried, or
You can find other AIs that does the web searching for you.
Hope this helps! Let me know if there are any further questions or issues.
— @samliew
Hello @wiro; how live of data are we talking?
An alternative approach for how we handle our conversational chatbot is to load information into an OpenAI vector store that’s associated with an OpenAI Assistant. Then we query that assistant via Slack.
Each day, the vector store auto-updates for material changes, like updates or removals. Our trick is to convert divergent data sources into common Markdown and it’s that output which goes into the vector store.
The vector store file that’s saved has header, reference URL, and byline above and below the primary materials so that we can properly link to citations.
Another good alternative to web searching is Perplexity. It’s an AI that can scrape the web / summerize links that you give it. Perpelxity has its own module on Make
I’ve since written a compilation of web scraping strategies, take a look here
For more information on the different methods of web scraping, see Overview of Different Web Scraping Techniques in Make 🌐