What's the Vibe?
- Feb 6
- 5 min read
After what feels like years of hype, AI is here and it's magic. In fact any new technology that feels like magic proves there's been a jump in innovation.
If you're not sure even just going to https://gemini.google.com/app and asking "Tell me how the UK electricity system works" is a good demonstration. This is the response I got.
The UK’s electricity system is currently undergoing its most significant transformation since the post-war era. It has moved from a centralised, fossil-fuel-heavy model to a decentralised, low-carbon network managed by a new public body.
Here is a breakdown of how the system works in 2026.
The Core Components
The system is divided into four main "stages" that get power from the source to your socket:
Generation: Power is created by wind farms (offshore and onshore), solar parks, nuclear plants, and gas power stations. Notably, coal was completely phased out in 2024.
Transmission: High-voltage "electricity motorways" (pylons and undersea cables) move power across long distances. This network operates at 275kV and 400kV.
Distribution: Local "B-roads" (smaller substations and poles) step the voltage down to safely deliver it to homes and businesses.
Supply: These are the companies you pay (e.g., Octopus, British G&$). They don't usually own the wires; they buy electricity in the wholesale market and manage your billing.
Who Runs the Show?
The landscape changed significantly in late 2024 with the creation of a new central authority.
NESO: The National Energy System Operator. A publicly owned body that balances supply and demand every second and plans the future grid.
National Grid: Now primarily a Transmission Owner. They own and maintain the physical pylons and cables in England and Wales.
DNOs: Distribution Network Operators. Regional companies (like UK Power Networks) that manage the local wires in your area.
Ofgem: The government regulator that ensures prices are fair and companies follow the rules.
The Energy Mix (2025/26 Stats)
The UK is now one of the cleanest major economies in terms of power generation. In a typical month in 2025/26, the mix looks roughly like this:
Renewables (~45–50%): Wind is the "backbone," often providing over 35% of total power. Solar and biomass provide the rest.
Fossil Gas (~25–30%): Used as a "flexible friend" to fill gaps when the wind doesn't blow, though its use is declining.
Nuclear (~10–12%): Provides a steady "baseload" of carbon-free power.
Interconnectors (~10–15%): Giant undersea cables connecting the UK to France, Norway, Denmark, and Belgium, allowing us to trade power with Europe.
Balancing the Grid: The 50Hz Dance
The most complex part of the system is that electricity cannot yet be stored easily at scale. NESO must ensure that the amount of electricity being generated exactly matches the amount being used at every moment.
Frequency Control: The grid must stay at exactly 50Hz. If demand drops (e.g., everyone goes to bed) and generation stays high, the frequency rises. If demand spikes (e.g., a "TV pickup" during a major event), it falls.
The Balancing Mechanism: If the grid is out of sync, NESO pays generators to turn up, turn down, or asks large batteries to discharge in milliseconds to stabilise the system.
Interconnectors: If we have too much wind power, we export it to Europe. If we have a shortage, we import it from Norwegian hydro or French nuclear.
5. What’s Changing Right Now?
In 2026, the system is facing three major pressures:
The "Great Grid Upgrade": A massive build-out of new pylons and cables to connect huge offshore wind farms in the North Sea to the rest of the country.
Electrification: Demand is rising for the first time in years as millions of people switch to Electric Vehicles (EVs) and Heat Pumps.
Digitalisation: The grid is becoming "smart." Many people now use "Time of Use" tariffs to charge their cars at night when electricity is cheapest and the grid is least stressed.
Fun Fact: On April 1, 2025, the UK set a record by running on 97.7% zero-carbon electricity for a 30-minute period. The goal for 2026 is to achieve a full 24 hours of 100% zero-carbon operation.
I always check the sources as AI still has a tendency to make stuff up. These are the sources it gave me which I'm reasonably ok with:
Ember on renewable breakdown.
The Guardian on grid structure.
UK Government site on electricity networks infrastructure.
UK Government on the split of NESO and National Grid.
National Grid article on transformation.
Carbon Brief analysis on 2025.
And inevitably a Wikipedia article.
[I've corrected a few American spellings (predilection for Zs instead of Ss) but otherwise this is exactly as the AI wrote it.]
Vibe Coding
In software terms, AI has been given a new name - Vibe Coding - so I thought I'd give it a go by trying to create a basic energy dashboard using data from Octopus, NESO, National Grid, Open Meteo, the EPC register and the Carbon Intensity analysis.
I started with an AWS account and set up AWS Amplify to give me a basic hosting environment that could take care of servers, memory, user access and some security. Amplify gives you a basic To Do app to experiment with which helps get something started - my experiment is to try to do all this without writing a single line of code.
The data sources, including our own, all have open APIs so I wanted to try and get a range of interesting data points and show these as current state of my home and the grid. Each API specification is quite different so I needed code that could connect to each API, store any credentials (such as my Octopus API Key), and present tile style dashboard for each.
Using VSCode I have access to the copilot and chose claude sonnet 4.5. Then it's as easy as asking a natural question as per this screenshot in which I've asked it to connect to our open API.
In less than a minute it's written 349 lines of code and I've got the core integration to the Octopus API up and working. It took a bit more back and forth asking for the entry boxes to add my own A-xxx and API key, then asking to find the smart meter half hour import and export data and then creating a graph of the data. I went a step further and asked for the latest Home Mini data point too and to display that as a dashboard tile.
And then connecting up a few more sources and more experimentation prompting to create a grid tile layout and getting creative with an interesting dashboard tile design I've got a laive dashboard of interesting datapoints. The AI came up with the idea of tilting the tiles when hovering over them (I have no idea why) and clicking takes me to a config/set up page for each source of data.
It also decided to implement automatic updates every couple of minutes without me asking which makes sense as this is a prototype live dashboard.
I've not prompted for a specific architecture so wondered where my Octopus API credentials are stored. Asking it and I discover they're placed in a browser cookie so not secure at all. Knowing this I was able to ask it to implement a more secure method which it then carried on and completed.
So like any tool (if you follow me on Facebook you'll know one of my hobbies is hedge laying which involves using a chainsaw) knowing how to use a tool and where appropriate being trained on it is key.
Comments