Getting started with AutoQL can be broken out into four key stages: connect, train, implement, and go live.
Stage 1 – Connect:
We’ll work with you and your team to configure an encrypted, read-only connection between our cloud and your database. Without storing your data, this connection allows us to train a custom model, execute queries, and pass data back to you. We use thorough security practices to make sure none of your users’ data ends up where it shouldn’t. You’ll be able to configure access and permissions for individual customers and users later on.
Stage 2 – Train:
Next, we get to work generating training data and building a custom language model specific to your database. We begin with unsupervised machine training based on our system’s current understanding of typical database structures, allowing us to tailor our model to your app and deliver a feature that your users can operate intuitively. Semi-automated training is also available if you wish to customize certain queries to fit terms or jargon that are unique to your business.
Stage 3 – Implement:
Once your custom language model is ready, it’ll be your turn to build and implement a conversational UI in your solution, letting your users query their data from anywhere in your application. At this stage, you can make use of any of our flexible implementation options (all widget components are open source) or opt to build directly on top of our API.
Stage 4 – Go Live:
With AutoQL trained on your database and implemented in your interface, you’re ready to start offering conversational access to data on demand to users! As users ask questions, their conversational English will be sent up to our cloud and translated into a query language appropriate for your database. Relevant data is then returned to them in easy-to-understand visualizations, all within seconds.