Member-only story
Reimagining the Recommendation Engine
Demonstrating The Utility of AI Assistants as a New Flavour of Recommendation Engine
The Rise of the AI Assistant
AI is transforming the way in which we interact with online apps. At present, we navigate them using search engines, news feeds, and menus meticulously designed to guide us to the information or service we require.
However, I believe the current innovations in AI make a new type of recommendation engine possible: the AI assistant.
We have already seen evidence of this with the emergence of Chat-GPT and Bard. However, let’s consider an example to illustrate this point further.
Imagine that you’re vacationing in London, and you wish to meet some friends for a picnic in Greenwich Park.
Before you decide on the date, you’d probably find your preferred weather forecasting service, maybe via a Google search or directly.
Next, you’d navigate their site and examine the weather forecast across your chosen date range, selecting a day and time according to the most favorable weather.
Wouldn’t it be easier if you could just simply ask? “What’s the best day to have a picnic in Greenwich Park?”.
No need to scour Google, no need to select your day. If it’s linked to a voice assistant, there might be no need to even type anything, you just receive a recommendation from your trusty AI weather assistant telling you the best date to arrange your trip.
In this article, I demonstrate how this is possible using function calling from Open AI, Langchain’s SQL Database Chain, and the Open Weather Map API.
Pay attention to the Gifs below showcasing the prototype I developed in action.
The first shows the response to “What’s the best day to have a picnic in Greenwich park?”
