I happened to notice Notion publicly started alpha tests of their new AI integration in this missive by Swyx on his L-Spaces substack. They appear to be really hyping this integration.
Unfortunately, as Swyx points out, the first encounter of Notion’s AI features in the wild produced some unfavorable outcomes starting with a complete reverse engineering of the GP3 prompts that Notion features utilize.
This article by Swyx is an interesting dive into some of the deep regions of OpenAI’s system and perhaps suggests that Notion may have jumped the gun on their AI strategy. What are your thoughts?
Coda has a killer AI strategy; we just don’t know it [yet]. Care to elaborate Codans?
I have access to it and it is great, IMO the OpenAI Pack is not enough and Coda has to implement a deep integration as a core of the product, similar to Notion where your prompts directly modify the text in the canvas and not just a formula that returns a result.
There is a lot of value to be unlock!
Spot on! Could not agree more.
This has been a long-standing issue that I have overcome with my team using this approach.
However, while Text Blaze certainly provides us with direct canvas insertions and assessments, it is not a general approach Coda users should be forced into to accelerate content production and productivity. The text canvas must be addressable and support event handlers that can trigger Pack executions. Many users have said exactly this.
Oddly, there is silence from the Codans concerning the AI roadmap, which suggests they’re on to something big or they’ve realized the risks are too big to climb on the OpenAI bandwagon.