How EDF is making the most of its data with Snowflake

Alex Read, senior enterprise product manager for data at energy provider EDF UK, is charged with helping his organization make the most of its information assets. Perhaps unsurprisingly, the hottest technology topic and the issue his bosses raise the most, is exploiting AI.

“It’s at a point where there are visions at an executive level of where AI is infused throughout our technical services and capabilities,” he says. “Our job as data and technology experts is to understand where we can deliver the most value, and also make sure that what we're doing, from an enablement perspective, is designed to be adopted and operationalized at scale.”

Read says it’s a responsibility he relishes, describing the demands of his role as fast and varied. As part of the enterprise IT function, Read helps over 1,000 users across the business to make the most of the EDF UK’s data tools. His team sits at the heart of a federated hub-and-spoke model, where they focus on enablement and business units concentrate on value.

“I'm accountable for the tooling, architecture, and common services-type capabilities in data and analytics,” he says. “Then in the federated business units, they're on the hook for data products and value extraction. They have teams of data engineers, data scientists, and ML Ops engineers, who are building data, AI, and ML products that create value for EDF UK.”

Latest Videos From

Read adds the federated approach gives his business “the best of both worlds” – one data platform across EDF UK that creates a consistent approach to information, and product teams in organizational functions that are naturally closer to business value.

“That's the sweet spot we've found,” says Read, who explains how the Snowflake AI Data Cloud is the heart of his organization’s data approach. The platform provides a single, consistent one-stop shop for hundreds of potential data sources.

“The data needs to be available at a cadence that can be used to operationalize near-real-time use cases,” he says. “So, we've got things like dynamic tables and event-driven architectures that enable that data to be brought onto the platform.”

Snowflake forms part of an integrated data stack that includes dbt, Matillion, Amazon Bedrock, and Sagemaker. Read says people across the organization can use the tools within the stack depending on their role, function, and projects: “Data stacks serve multiple personas – that’s the easiest way to describe it.”

With solid foundations in place, Read’s team has started enabling the structures that allow the rest of the business to exploit AI and ML. The next step is to ensure enterprise data is high-quality and well-defined so that emerging technologies know its context.

“We’ve got Snowflake services we're using, like Semantic Tables and Horizon Data Catalogs, whereby we're defining our data assets and making sure they're understandable to AI agents,” he says. “And then, we're building capabilities via Cortex Analyst and Cortex Code on top of that platform.”

The results of this construction process are becoming visible across the organization. The data platform already supports data initiatives across EDF UK’s Retail, Business, Wholesale Markets, Finance, and Zero Carbon Homes divisions. In Wholesale Markets, the company is rebuilding its volume forecasting platform, the analytics engine behind purchasing and hedging decisions in an energy market worth more than $13.53 billion (£10 billion), on Snowflake.

“That's about taking data from industry flows, weather providers, and internal retail and business solution consumption patterns,” he says. “We're essentially building forecasts in AI models, whereby we will be able to forecast the amount of energy required across the market. As these are such big numbers, even small percentage improvements in the margin of error and your ability to forecast, price, and hedge can create significant financial benefits.”

While those data-enabled efforts take place behind the scenes, Read’s team is also pursuing significant customer-facing initiatives. One major driver for these initiatives is the Market-Wide Half-Hourly Settlement, a program of industry reforms led by energy regulator Ofgem, which aims to provide a more accurate picture of electricity consumption across the UK.

Read says the regulations have inspired EDF UK to engage with data and explore innovative billing techniques. FreePhase, for instance, offers three pricing bands that reward customers who shift their usage to lower-cost periods. “You can't create those kinds of products without the ability to access data at a fine granularity,” he says.

Another example of an innovative billing technique is the Sunday Saver Challenge, EDF UK’s smart meter-enabled program that rewards customers with up to 16 hours of free electricity on Sundays when they shift weekday energy use from peak demand periods, easing pressure on the grid.

Customers contact the firm with questions about the challenge, such as its rules and benefits. Now, Read is using agentic AI to help service center staff deal with these queries effectively. His team is developing an agent using Snowflake Cortex technology that pulls information from the data platform and presents insight in the staff’s Slack user interface.

“The agent is being built as we speak, and is pretty close to being productionized on Cortex,” he says. “That agent will automate a lot of the questions being answered by our energy specialists. A customer query will be automated against Snowflake. So, essentially, they’ll be talking to the data on the platform to answer questions.”

At a time when many businesses struggle to turn agentic explorations into production services, the developments from Read’s team demonstrate the potential power of AI to boost internal operations and customer services. The key lesson for other digital leaders is to spend time on delivering the right outcomes.

“Snowflake comes highly integrated out of the box, and a minimal number of vendors are required to get to value,” he says. “And that's what our focus has been, and that has been the key lesson from this transformation – remove any barriers that get in the way of delivering value quickly.”

The result is that a business eager to reap the benefits of AI-enabled innovation is beginning to adopt and operationalize data-powered services at scale. Read reflects on progress so far and paints a picture of data and AI-enabled services at EDF UK in the future.

“Snowflake will be serving near real-time use cases via its platform with data that is of acceptable quality, defined, catalogued, and accessible to AI models, which have touchpoints out into the wider operation,” he says. “People of all skill sets and personas will have AI tools in their hands, fed by Snowflake. I think that's what good looks like for us.”