How to Implement Autonomous Optimization Projects Using Amp.ai

Olcan Sercinoglu

Sep 10, 2018

In a previous post we looked at how to design effective Autonomous Optimization (AO) projects. In this post, we describe how to implement our AO projects using Amp.ai.

Create Amp.ai Project

We start by logging into a new Amp.ai account, where we are greeted by Amp.ai bot (lower left) and Nik (lower right), an AO expert from Scaled Inference. Amp.ai bot has already created our first project for us and is ready to walk us through the remaining steps using the red dot.

ao-impl-create.png#asset:705

Integrate Application with Amp.ai

Next, we choose our type of application and follow the instructions to connect our application to our Amp.ai project. For web (Javascript) integrations, we simply drop an HTML tag into the head section of our web pages.

ao-impl-integrate.png#asset:704


We need to work on our integration until the received sessions reflect our design in terms of the outcome events for our metrics, decision events for our variants, and context events for our segments. We illustrate this below for our Hello World example web app.

ao-impl-integrate-check.png#asset:702


Amp.ai provides several features to help ensure correct implementation, such as Recent Sessions, Sample Sessions, Session Summary, and Events Summary.

ao-impl-features.png#asset:718

Start Autonomous Optimization

Finally, we define our metrics, approve our decision points, and allocate a percentage of our sessions to be autonomously optimized (or amped).

ao-impl-start.png#asset:703

In future posts we will dig deep into specific use cases across the customer funnel and software stack.