Many tools will enable this in the coming months and years. Only a few will be killer apps. The reasons are many, but all businesses should begin with the underlying objective;
… making domain knowledge part of the AGI experience.
Canolli or Cake explores this topic and suggests that GPT Plugins may not be the ideal pathway for users to converse with their business data.
In the context of CustomGPT, the nature of your SaaS model provides significant SMB and enterprise value; plugins are simply one avenue for knowledge work (more here). I believe many more avenues exist that embrace knowledge work and where AGI can move the needle for greater productivity and output.
CustomGPT baked into platform workflows will have far greater demand than GPT Plugins from organizations that can afford to compress the time it takes for their knowledge to become “chatable”. In my view, the copilot industry is about to explode.
Google Apps Script - an integration library that makes it easy to identify content, build the training corpus, and and deploy ChatGPT projects instantly.
Google Workspaces - integrated sidebar UIs that draw ChatGPT services closer to knowledge work. (ex. below) Imagine a Google Slides plugin that generates a presentation based on a general topic and access to all your marketing and product content.
No-Code Database Extensions - products like Airtable and other closed systems that LMMs cannot be trained on will have ravenous demand for real-time data-to-chat interface demand.
Document Platforms - such as Coda can serve as corpus-to-chat functionality in near instant production-ready services.
I’m researching a very long list of CustomGPT “plugins” related to actual business workflows and productivity. I think this is where companies will spend serious amounts of cash to put AGI into the hands of its workers. I don’t think businesses will encourage workers to use ChatGPT as a primary interface to work with their own business data.
Thanks Bill - 100% agree. The business case for plugins is still up in the air, since the “App Store” needs to shape out. If it indeed becomes bigger than the App store, then that will certainly make for interesting times.
Till then, I agree with you – just the use-cases emerging with the CustomGPT API have far reaching applicability, and we are racing to make more features of the API widely available.