April 10, 2024
2 minutes read
The other week we needed to build a solution to enhance notes for a knowledge management system. For this project we used Tape and OpenAI’s GPT API.
As always you must be comfortable with what data you are sending to online services.
The first thing we did was on a note creation we sent the note to the GPT API along with a list of ‘Note Types’ and our definition of an ‘Atomic Note’ and requested four things:
- That the note be corrected for spelling mistakes, grammar and clarity.
- That the note be categorised into one of the ‘Note Types’ we had provided.
- That the AI should assess the note against our definition of ‘Atomic’ and flag if the note does not match that definition.
- Finally we ask for a Note Title.
We use the data that is sent back to update our note immediately enhancing the ‘quality’ of our notes.
The next step is to automatically categorise (file) the note, it doesn’t matter if you use P.A.R.A, Johnny.Decimal or any other system, in this project a variation on Johnny.Decimal was used. The system was tracked in Tape already so we pulled the data from existing records and sent it up to the AI along with the note requesting the relevant categories be assigned.
It should be remembered here that the AI can only ‘file’ a note based on its contents and the system details it is sent rather than the ‘intention’ of the note, using this method with documents or emails has a much higher success rate
On our Note record we have a few buttons:
If we click the ‘AI Idea Growth’ button we ask the AI to expand on the note and generate more ‘Atomic’ notes based on the original when we get those back we create a series of new notes and link them all together.
This is all very useful and adaptable to several areas including ‘Shared Mailboxes’ or even incoming invoices. However, GPT shines when you can have a ‘conversation’ with it to expand on ideas and dive off into other areas in the form of ‘conversational learning’ and notes are a prime area for this.
To solve this we use Tape’s record comments area and when a user prefixes a comment with ‘AI’ (this could be anything) it triggers an automation which sends the comment off to the GPT and then posts the response back into the record comments:
The system builds a ‘memory’ of the current chat as it goes:
Imagine a scenario where your team can effortlessly collaborate with an AI assistant, seamlessly correcting errors, organising data, and brainstorming ideas in real-time.
By harnessing the combined power of Tape and OpenAI, businesses can gain a competitive edge in today’s data-driven economy. Whether you’re a small startup or a multinational corporation, integrations like this offer unparalleled opportunities for growth and success.
Reach out to us today and discover how integrating Tape with OpenAI can transform your business and maximise efficiency like never before.
