As generative AI evolves, certain trends are becoming clearer,
In yet another milestone in AI consulting giant McKinsey unveiled its own generative AI tool for employees called lilli
a) McKinsey launching this agent gives credibility to the domain for enterprise AI assistants
b) On one hand, it’s a familiar copilot strategy – but it also points to a shape of things to come.
I have never been a fan of the end user chatbot use case. However, the AI assistant for the experts within the enterprise is a perfect use case (something that the AI critics consistently miss)
c) The technology is LLM agnostic – LLMs themselves will no longer be a differentiator
This development is likely to fuel a renewed interest in AI automation – potentially completed by technologies like knowledge graphs.
There are some additional improvements to the chatbot technology.
1. Providing Context. combine proprietary knowledge with general knowledge in a useful way.
2. Cost. Reducing the “cost per query” down a hundred-fold between the first pilot and the launch, and down again four-fold since then.
3. Confidentiality. How to keep internal knowledge private
4. Confidence in result. How to show the sources of answers.
I believe that this type of chatbot is the shape of things to come i.e. an agent that complements the employees (and is not customer facing)
In the following part of this post, we will discuss why the consumer use case is not proven.
Image source and reference: