From Signals to Strategy: Horizon Scanning for Energy Network Innovation
The UK energy sector invests heavily in network innovation, yet much of what it learns is lost before the next project begins. Fourteen GB network licensees run their own innovation portfolios, funded through NIA, SIF and RIIO allowances, and each one is working to deliver against demanding commitments on net zero, connections and network resilience.
The knowledge produced by that investment is spread across closedown reports, DSAP papers, working group outputs, business plans and the memories of the people who ran the projects. Once a project closes, its report is published and archived, and the lessons it contains rarely shape the next round of investment. When people move roles, a great deal of that context leaves with them.
The result is that the same use cases, such as demand forecasting, asset health, LV visibility and customer vulnerability, are investigated several times over by different networks with little shared learning. Strategies are written carefully, but they are often out of date within a year because the technology landscape moves faster than the planning cycle. Horizon scanning, done well, is one of the most practical ways to close that gap.
What horizon scanning is, and why it matters now
Horizon scanning is the structured, ongoing search for early signs of change that could affect an organisation, such as emerging technologies, new regulation, shifts in policy, or developments in other markets. In a network business it could be used by heads of innovation, innovation project leads, funding managers and strategy teams to decide which ideas deserve investigation and which can safely wait.
The timing makes this more important than usual. RIIO-3 is under way, ED3 business planning is approaching for the distribution networks, and the connections queue has become a national issue that affects investment and economic growth. At the same time, AI, sensing, flexibility markets and digital twin technologies are maturing quickly, and Ofgem is creating new routes such as its AI sandbox for testing these technologies in a regulated setting.
In that environment, an innovation team that sees a relevant development six months earlier can shape a funding bid, avoid a duplicate investigation, or bring a proven approach in from another network or country. The value of horizon scanning lies less in spotting technologies and more in recognising, early and with confidence, which of them matter to the objectives the business has already committed to.
Where traditional horizon scanning falls short:
Most horizon scanning in the sector still takes the form of a periodic report or a technology radar, compiled by a small team or an external adviser and refreshed once or twice a year. These outputs are often well researched, but they tend to share a handful of weaknesses that limit their influence on real decisions.
They are disconnected from strategy. A list of promising technologies does not tell an innovation manager which business plan commitment, regulatory outcome or operational problem each one could help with.
They ignore what the organisation already knows. Lessons from previous innovation projects, and the ideas that colleagues and partners submit through internal idea platforms, are rarely brought into the same view as external signals.
They date quickly. A document produced in spring can be overtaken by events by the autumn, and there is seldom a mechanism to tell readers which findings have changed.
They lack provenance. Readers often cannot see where a claim came from, how many independent sources support it, or how confident the authors were, which makes it hard to defend a funding decision to a regulator or an investment committee.
None of these weaknesses reflects a lack of effort from the people doing the work. They are the natural limits of a manual, document-based process trying to keep pace with a fast-moving field.
What energy can borrow from financial services
Financial services firms have faced a similar problem for many years in the form of emerging risk. Geopolitical events, regulatory change, supply chain failures and concentration in critical third parties can all affect a firm's resilience, and regulators expect firms to show that they identify these risks early and act on them in a controlled way.
Horizon scanning is an established practice in financial services, where banks, insurers and market infrastructure providers monitor emerging risk continuously rather than once a year. Mature approaches combine specialist research providers, newswires and curated news sources with AI agents that read, classify and summarise what they find. Each development is mapped into a knowledge graph that connects events, regions, sectors, regulations and the firm's own exposures, and analysts receive structured briefings and alert feeds that they can question in plain English.
Several practices from financial services translate directly to energy innovation.
Curated sources over open search. The quality of the output depends on deciding in advance which sources are trusted, and treating everything else as lower confidence until corroborated.
Structure before summary. Each signal is mapped into a defined model of entities and relationships before any summary is written, so that findings can be compared, filtered and traced over time.
Provenance on every claim. Every briefing links back to the documents it was drawn from, which allows a specialist to check the reasoning rather than take it on trust.
A human review gate. Agents propose and people decide, and nothing is escalated or published without a named reviewer agreeing that it should be.
Continuous rather than periodic. Monitoring runs every day, so the picture of risk is always current and changes can be seen as they happen.
Swap emerging risk for emerging technology and regulatory change, and swap the risk register for the innovation portfolio, and the same pattern gives an energy network a horizon scanning capability that regulated organisations already rely on elsewhere.
Mapping signals to strategic objectives
The most valuable change is to connect every signal to the objectives the business has already committed to, so that horizon scanning answers the question of why a development matters as well as what it is. A context graph is the practical way to do this, because it holds the relationships between signals, technologies, initiatives and objectives explicitly rather than leaving them implied in prose.
A context graph extends a conventional knowledge graph by recording, on every node and relationship, where the information came from, when it was observed and how confident the system is in it. A claim drawn from one trade article sits at lower confidence than the same claim corroborated by a standards body and an academic trial, and the graph makes that difference visible to the reader.
The chain of connections typically runs as follows.
External signals such as a trial result, a vendor announcement, a new standard or an Ofgem consultation are captured from curated sources.
Technologies and capabilities are identified within those signals and assessed for maturity, using a shared scale such as technology readiness levels.
Business initiatives are linked where a technology could help, drawing on innovation projects, business plan commitments and operational programmes.
Strategic objectives sit at the top, taken from the DSAP, the RIIO-3 or ED3 business plan, sustainability action plans and the regulatory outcomes the network reports against.
Two internal sources make this chain far richer. The first is the archive of previous innovation projects, including closedown reports and lessons learned, which shows where similar ideas have been tried, what barriers they met and what conditions led to success. The second is the stream of ideas submitted by colleagues and partners through internal idea platforms, which can be matched against external evidence to show which suggestions are supported by developments elsewhere.
With those connections in place, an innovation manager can ask which emerging technologies could advance a specific objective, which past projects have already explored them, and which internal ideas they support. The system can also raise an alert when a new signal touches an objective that is falling behind, so that attention goes where it is most needed.
A practical method for your innovation team
If you are responsible for innovation in a network business, you can build this capability in three stages, each of which produces something useful on its own.
Learn from what has already been done
Start by gathering your own innovation history, including closedown reports, project lessons learned, funding bids and the strategy documents that shaped them. Use AI to extract the technologies, barriers, outcomes and people involved, and map them into your graph so you can see which themes recur, why promising projects stalled, and which conditions were present when something moved into business as usual. This step alone often reveals duplicated effort and forgotten work that deserves a second look.
Scan the horizon
Agree a set of trusted external sources with your specialists, such as Ofgem and DESNZ publications, ENA and Smarter Networks Portal material, academic research, international programmes such as EPRI and ENTSO-E, and selected industry press. Configure agents to monitor these sources continuously, summarise what is new, assess each technology for maturity and applicability to your network, and record the source and confidence of every finding. Keep a human review step so that your specialists decide what is escalated and what is filed for later.
Shape what comes next
Link the signals and lessons to your strategic objectives and live initiatives, and use the combined view to draft and test your innovation strategy. You should be able to show, for each priority, the evidence behind it, the past work it builds on and the external developments that support it. Because the strategy is drawn from the graph, it can be refreshed as the evidence changes rather than rewritten from scratch each cycle.
Keep the loop running
Treat the capability as a cycle in which you discover new material, ingest it with provenance, model it into the graph, reason over it to support decisions, and feed the outcomes of those decisions back in. Review the source list and the objective mapping each quarter, so the system continues to reflect what the business is trying to achieve.
From one network to the whole sector
A single network that builds this capability will make better innovation decisions, but the larger prize lies in sharing it across the sector. Many of the signals that matter to one licensee matter equally to the other thirteen, and the lessons held in one network's closedown reports could save another network a year of rediscovery.
A shared horizon scanning capability could sit within a sector AI Centre of Excellence, built on a common ontology so that every member describes technologies, use cases and outcomes in the same terms. A federated design allows each network to keep control of commercially sensitive data while contributing to, and drawing from, a shared view of what the sector is exploring, what has worked and what has stalled.
Other industries show that this kind of collaboration can work when the boundaries are clear. Pharmaceutical consortia have used federated learning so that partners share models without sharing raw data, and pre-competitive alliances in life sciences have drawn a clear line at the outset between what is shared and what each member competes on. For energy networks, the same principles would let regulators see a coherent sector picture instead of fragments, and would allow a use case proven by one network to be adopted by others far more quickly.
Getting started
You do not need a large programme to prove the value of this approach. A focused prototype of around four weeks, using your real strategic objectives, a sample of your innovation archive and a defined set of external sources, is usually enough to show whether the insights are useful and what it would take to run the capability in production.
A few principles will help that first step succeed.
Agree the specification, the data model and the integration points before any build begins, so that the prototype reflects how your business actually works.
Involve your innovation leads throughout, with regular working sessions that use real scenarios, so that the people who will rely on the system shape it as it is built.
Use public and already approved data sources first, and bring in sensitive internal data once the governance is agreed.
Plan for scale from the outset by assessing the infrastructure, security, data and support needed to move from prototype to an operational service.
Horizon scanning that is connected to strategy turns a stream of interesting news into decisions that a network can explain and defend. How quickly a network can get there depends on its wider AI maturity, including the state of its data, the readiness of its people and the rigour of its engineering, all of which vary considerably across the sector.
Every network operator is now working with AI, but far fewer can say with confidence where their projects stand. Our sector benchmark report, AI Maturity in UK Electricity Networks, publishing in October 2026, draws on structured interviews with directors and heads of data, engineering, regulatory and strategy functions across transmission, distribution and NESO, supported by a sector-wide survey.
It examines live projects, planned investment including Ofgem approved AI spend, data foundations, workforce readiness and engineering standards, and places each participating organisation on the five-stage Deliberate AI Maturity Framework.
Register your interest to receive the report ahead of its general release. If you would like to explore what AI powered horizon scanning could look like for your organisation in the meantime, we would be glad to talk at hello@webuild-ai.com.
