Posted on Dec 17th, 2019 by James
Ok admittedly I might have bored you with this one before but hear me out. Google have been working on the flutter framework for a very long time and in more recent times it has started to pick up real pace of adoption.
Based on the Dart language, in recent times it has been mentioned in the trial section of the ThoughtWorks Tech Radar and in the GitHub Octoverse report Dart ranked number 1 as the fasted growing language and Flutter ranked number 2 for fastest growing open source project.
With its latest release that brings their Flutter for Web into Beta status it means that developers are nearing a world where they can write one code base and truly target multiple devices - mobile, web and even native desktop. A term google call Ambient Computing.
In 2020, I think we'll start to see the ramp of adoption and that initial approach will be from organisations looking at replacements for cross platform mobile frameworks such as the popular JS based react-native framework.
Serverless is not anything new. Lots of organisations have started to adopt serverless technologies whether its AWS Lambda, GCP Cloud Functions, Azure functions or any other type of serverless offering are being deployed into production.
This is because the operability capabilities of serverless are starting to grow - as organisations utilise serverless in production they are experiencing the battles and sharing the war stories of making those functions production ready - monitoring, automation, logging, security etc. There are also a number of observability tools such as Honeycomb, OpenTracing, Epsagon and Dashbird specifically targeting serverless environments.
As the trend of increasing operational readiness in serverless continues I think we'll see further adoption.
Machine learning and this whole AI bubble isn't going away but I think how we access it and begin to utilise it is changing. AI was originally for those data scientists of the world, those people that understood various models and could apply those models to datasets using the language of choice. Have you ever wrote a Jupyter Notebook? Yeah me neither.
The power of machine learning (whilst not without its problems - a quick google on AI bias shows this ) is certainly not going away but there is an intermittent step.
A step where humans can build relatively straightforward models without having to know too much of the underlying structure. AutoML and AutoML (Tables) are examples of machine learning made accessible - a trend I think both enterprise and start-ups will look at adopting over 2020 as they step towards full AI and machine learning.