Bespoke Experiences aka Recursive Ephemeral Multimodal Generative Interfaces
What if we didn’t organize our data into list views and record views?
Maybe the front-end interface is now going to reflect a new type of computational back-end?
The same way that static software as a service atop a relational database made sense, perhaps a recursive ephemeral multi modal generative interface make sense on top of a large language model.
Let’s break down the concept:
recursive - meaning moving iteratively towards some state or objective
ephemeral - meaning short lived or impermanent state changes
multimodal - voice, text, vision
generative - leveraging AI to predict the next character, pixel, byte
interface - a way to to interact with a piece of technology
In Google Gemini’s Bespoke Experience video…
We saw an example of this concept.
Could we extend this to be applicable to business software in the traditional sense of SaaS.
Could we use generative AI to autonomously render react components?
We ask a question against our system
Get back a response using RAG
Dynamically generate new components in the interface based on the data
As more input data comes into the system the next message is rendered with the new chart. This concept is not new in social media news feeds, ad tech services etc, but in enterprise software most of the interfaces we use are static; when it could be and should be always predicting the next best user interface for the end user.
Generative charts using Tremor.so.
We can provide chat examples so the AI has an idea of how to use the library.