Getting started with AI Context Engine
The data.world team will work closely with your team to prepare and set up the AI Context Engine.
The setup involves some prep tasks, such as identifying the questions you want answered, virtualizing the data that will answer the questions, and preparing metadata, most importantly, setting up the Business glossary that will be relevant to the questions.
Next, in the Knowledge Capture phase, the data.world team will work closely with you to analyze the questions, and come up with the ontology and mappings.
Finally, in the Knowledge implementation phase, the data.world team will create the ontology and R2ML mappings. The team will test your questions against these assets and will deliver the required assets to you through a project in the data.world catalog. Using the data.world built Developer Chat Tool, or any other tool of your choice, you get to chat with your data and take the AI Context Engine for a test drive.
Start Small
When you begin setting up AI Context Engine, for the first iteration, start small with a small set of business questions you want the AI Context Engine to answer and break it down into even simpler smaller questions to understand the process of setting up the AI Context Engine.
As for the team, start with a small team of three experts: Knowledge Engineer, Data Expert, SME. A smaller team helps stay focused and achieve what we need to. Most of the work is done by the Knowledge Engineer in consultation with the Data expert, and the SME. After the first iteration is completed successfully, we should get results and get people excited. In a subsequent iteration, we can bring in a few more people.
Prepare your questions
Define three business questions in English that you would like to answer, including:
Question: What business question do you want answered?
Example: How many people between the ages of 15-45 booked a cruise with us in the last 2 years?
Persona: Who is asking the question?
Example: Marketing team, Sales team.
Motivation: Why are they asking this question?
Example: Goal to target specific demographic while designing cruises.
Baseline: How is this question being answered today?
Example: Provide any code/SQL query that is being used. Example Share any tables or SQL queries you currently maintain in your warehouse or elsewhere that may track this. If this is the first time setting up the AI Context Engine, these questions should start from simple and moderately increase in complexity.