The goal of this talk is to share the ups and downs of building a visual testing framework, on top of AWS Lambda and Capybara. I will share the challenges we faced and eventually overcame to have continuously integrated visual tests. When we started on this journey we hadn’t had any experience with visual regression tests. Our approach to visually testing out product before the start of this project was risk-based and fairly manual. Before starting to build this framework, we looked at solutions that would suit our needs and purposes. All of them looked good on their face but they either lacked depth (like missing cross-browser support) or they were too expensive to run. For these reasons, we built our own tool to help developers ship visually tested product without ever having to wonder if it is broken on a particular browser. When we started building this visual testing framework almost a year ago, we hadn’t the faintest idea of how challenging this would be (CSS animations, loading bars, images, fonts …). We had to implement hacks aka ingenious solutions so solve many tricky issues and flakiness. The experience building this framework has taught us lessons not only about visual testing but also about our products and what things we can do to improve the end user experience. This talk will cover all of these Trials and Triumphs in as much detail as time allows. Draft Outline Reasons for visual testing Currently solutions in the market Current visual testing architecture Trials/Triumphs: Image loading Trials/Triumphs: Fonts Trials/Triumphs: Animations Trials/Triumphs: Continous integration.