written by
Dries De Rydt

Google Cloud Next ’18 London — It’s serverless all the way

Events Artificial Intelligence 3 min read
The keynote room — Ready for 8000 GCP developers, sales reps, etc.
The keynote room — Ready for 8000 GCP developers, sales reps, etc.

On October 10–11 2018, we were invited to Google Cloud Next ’18 in London to find out what’s next for Google Cloud Platform. Following the earlier event in San Fransisco, we were wondering what news they would have for us now. The message was clear: Google is going all-in on serverless.

Google Cloud Next ’18 London
Google Cloud Next ’18 London

Overall atmosphere

The conference was packed: attendance has doubled since last year and it shows! All around there were presentation rooms and each was scheduled to host several interesting breakout sessions on all kinds of topics on a first-come-first-served basis and the queues were huge. Luckily it never felt too bad missing a talk, because there was so much else to do.

The convention floor was filled with Google Partners showcasing their products built upon or collaborating with GCP, as well as Googlers offering advice and explaining what the platform could mean for your business. These informal talks about Google tech are the biggest value a developer can get from conventions like these, so we invested our time heavily on these talks.

And if all that doesn’t appeal to you, there were always opportunities to hunt for snacks and coffee. Oh and some table soccer as well for the enthusiasts like ourselves.


Serverless was part of almost every presentation. Google is going to be expanding their offering by allowing you to use Docker for Cloud Functions. This means you are no longer limited to supported programming languages or toolkits! Need something to be installed on the OS for your code to run? Just wrap it into a Dockerfile and deploy it!

Google also talked extensively about Knative. This is a Kubernetes add-on that allows users to run serverless workloads on their own Kubernetes instance. In the current Kubernetes environment, scaling to 0 is not an option for any deployment, so this allows you to use your cluster more efficiently!

“The core of Craftworkz has always been prototyping, in a fail-fast and rapid development approach.”

Machine Learning

AutoML is a very interesting tool. There were demo cases of people who built classifiers without any coding experience. It’s interesting to think about the future of AI consultancy, and how far automated transfer learning can go.

Perhaps in a few years there will be no more demand for AI consultants, since companies are empowered by intuitive but versatile tooling.

The core of Craftworkz has always been prototyping, in a fail-fast and rapid development approach. AutoML is a great tool for us to quickly evaluate the merit of an idea, without having to start the long process of iterating over neural network architectures, evaluating and such.

They also announced BigQuery ML. It’s a new way to train and use ML models purely through the BigQuery SQL language. It supports simple regression models, meaning you could use it to easily predict missing values in your database without having to set up data pipelines and such. It’s another way for us to quickly get insights in order to determine the value in our (customers) data.

“No convention should ever be organised without a proper afterparty”

The Google Startup Afterparty

No convention should ever be organised without a proper afterparty, certainly not when you invite a company like ours, where the team has a mean age of about 27 years old. Google certainly more than filled our expectations by inviting us to the Dinerama venue in central London, a food market filled with young developers and startups for the occasion.

Dinerama — The venue for the Google Cloud Next Startup Afterparty
Dinerama — The venue for the Google Cloud Next Startup Afterparty

Next to having great food and drinks, the place was filled with relaxed developers eager to share ideas and Google Cloud experiences. These talks give a very nice perspective on your way of working en developing, while at the same time expanding your network and future business opportunities.


Google Cloud is moving towards a bright future and we are very eager to start prototyping with the tools they’ve built. Serverless is definitely the future for companies like ourselves who want to spend the most time developing production ready prototypes and applications that are ready to scale, and less time setting up environments and servers.

And while the future is serverless, Google still invests heavily in the tools for developers who want to be in control of everything they do, right up to the bare metal servers. But now you will be able to sprinkle some serverless Google fairy dust over it.

Want to read more about our stories, or about how we’re leveraging cloud providers such as Google to build our own conversational agent SaaS platform Oswald, please checkout our other blogposts here.

Events Cloud Artificial intelligence