CTO interview: Octopus Energy on using cloud power to get consumers to flip the switch on renewables

Training neural networks

The platform also provides Octopus with visibility of the energy consumption rates across its business by training neural networks using smart meter data that tracks how energy usage rates vary according to the time of day.

This in turn means the platform can make predictions about how usage patterns are likely to change throughout the day, and help the company decide the best time to acquire energy from the wholesale market to keep its customers’ energy costs down.

The latter concept forms part of the company’s Agile Octopus energy tariff, which provides customers with half-hourly pricing updates so they know when the most cost-effective times of day to ramp up their energy usage.

The insights gleaned from Kraken make this possible, while providing a level of dynamism, flexibility and choice to consumers regarding how and when they procure their energy in a way that few, if any, incumbent providers can match, says Eddison.

“The energy industry is based on technologies and processes that were probably designed in the 1980s and implemented in the 1990s, but it feels like it’s running on 1970s technology,” he says. “That’s why something as simple as switching your energy supplier takes a couple of weeks.”

But it need not be that way, says Eddison. “When we were very new into the market, I remember a time when the energy regulator Ofgem were consulting on whether we should move to faster switching, and some of the incumbents were saying there is no evidence customers want faster switching.

“We took the view that if there is no evidence that customers want faster switching, why not let us do it and we’ll see if customers like faster switching? Create the opportunity,” he says.

“We need to be able to have things where we can move, make changes faster and interact more flexibly”

What the company uncovered while digging into this is that switching energy suppliers is a process that could be carried out in a faster and far more cost-effective way if providers were willing to invest in their own technology stacks, says Eddison.

“The industry moves at the pace of the slowest, so what we’re doing with Kraken is we’re saying you shouldn’t need to have systems that are inflexible, change processes that take years and are incredibly expensive,” he says. “We need to be able to have things where we can move, make changes faster and interact more flexibly.”

Which is one of the reasons why the company has taken to licensing the Kraken platform to other utility providers, including E.On and Good Energy in the UK, as well as several other overseas suppliers.

“Kraken is much more than just a competitive advantage for us. It enables us to make changes to the way the market operates, so we can use data in real-time communications to really match consumption to energy generation, and progress on tackling climate change, while offering fairer pricing and more transparency [on their usage patterns] to customers.”

Licensing tech for growth

In order to grow its share of the market further, the company has embarked on a series of Kraken-focused licensing deals that seemingly serve to put its technology in the hands of its competitors.

While that might seem like an odd thing to do, it is a move that it hopes will lead to further consumer take-up of green energy tariffs across the industry, and – in turn – assist society in its fight against climate change.

“That wouldn’t work if you’ve only got one player in the market with access to that sort of technology, and if you look at who we’ve licensed it to in the UK, we’ve got Good Energy – who are a green energy pioneer, and E.On, which is the only one of the big six offering 100% renewable energy,” he says.

“It’s not a question of licensing it to somebody with the biggest pockets. It’s about licensing it to people who we think have the same values and objectives to then benefit from using Kraken.”

He says: “They buy into exactly the same model, so they get exactly the same enhancements and developments at the same time to the same extent that occurs with [the version of Kraken] we have.”

It is through these licensing agreements, coupled with direct engagement with customers via the Octopus brand, that the company is hoping to achieve its goal of having 100 million customers around the world relying on Kraken by the end of 2030.  “It’s a credible target, but there is still a lot of work to do to deliver on that,” says Eddison.

Cloud provider partner

Hitting its customer acquisition goal is also one of the reasons why the company has sought to forge even closer ties with its long-standing technology partner, AWS, in recent months, evidenced by its declaration in August 2020 that Amazon has now become its preferred cloud provider partner.

Kraken has been hosted in the AWS cloud since its inception, and the platform itself relies on several inter-connecting Amazon cloud services to run, with its Simple Storage Service (S3) used to store customer data, which is then processed by AWS Glue to prepare it for analysis by the Amazon Relational Database (RDS) offering.

“In old school tech, you would have invested in hardware in a datacentre, and had it sitting there 24 hours a day, seven days a week, and you’ve got finite [compute] capacity available to you,” he says.

“The AWS components are very good to work with… and having the ability to flex computational power to meet demand is really, really important and we heavily use that. And it means all we have to worry about is designing the application, and serving customers.”

Also, given the company’s want to expand globally through licensing agreements, being able to lean on Amazon’s global fleet of datacentres is another important factor in its decision to deepen its ties with the firm.

“We now have five Krakens live in production, spanning three countries, and those numbers are increasing. Each one is actually paired because we run a test environment alongside a production environment, and by the end of the year we’re going to have nine or 10 in use,” he says.

“We also don’t know which countries we’re going to be operating in in a couple of years time, and we don’t know what [data protection] rules are going to be in place within those countries either. So having a provider with a large geographical footprint and presence puts us in a strong position from that perspective.”

Furthermore, given the time-sensitive nature of energy pricing, being able to host Kraken as close to the consumers and providers that rely on it is also extremely important from a latency point of view.

And that is particularly important given the 50-strong team tasked with overseeing the development of Kraken are split between London and Australia at present.

“I have previously managed a development in Australia, with a technology team split between Australia and the UK, by coincidence, and the latency difference between running the application locally and running it on something from around the world is very, very noticeable,” says Eddison.

“If you want to do something which is highly responsive and performant, you need to move it closer to your point of use. So Amazon’s global footprint is definitely a key factor of the partnership.”

So with its race to reach 100 million customers by 2030 well and truly on, a growing priority for the company is to continue increasing its developer headcount to ensure Kraken continues to deliver on what it was designed to do.

“There are some exceptionally talented and dedicated people in the team, and we are growing rapidly, and looking for people who share the same passion for looking after customers,” he says.

“Have the team across technology and the customer-facing ops teams working together to try to make each other’s lives easier,” adds Eddison. “And that spirit of collaboration across the company is really precious to us, and a really valuable driver towards ensuring [Kraken] is efficient and easy to use.”