Data readiness, not AI hype
Ofwat published its first AI adoption plan for the water sector in June 2026. Read past the headline, and instead of it being a plan about artificial intelligence (AI), it’s one about data. Whether it can be trusted, who owns it, and whether a company can show its workings when a model produces an answer.
This distinction matters more than it sounds. Most of the sector’s AI conversation so far has been about use cases, from leakage detection and demand forecasting to customer service automation. Ofwat’s AI adoption plan is the first sign that the regulator is going to ask a different question first, one that shifts from “what is AI doing for you?” to “is the data underneath it something we can rely on?”
For water companies that have already started the data foundations work, it’s good news – and a test they can pass. For everyone else? If you ask us, it’s a deadline they didn’t know they had.
What Oftwat’s AI adoption plan actually asks for
Ofwat’s plan recognises that AI is already embedded across utility operations:
- Leakage detection
- Network optimisation,
- Customer service
- Billing
- Regulatory reporting
The regulator’s position has moved from whether AI should be used to how it’s deployed in ways that improve outcomes. That’s all while maintaining trust and accountability. Water companies remain accountable for decisions and outcomes that AI supports – the technology doesn’t dilute that responsibility.
The detail that should get the most attention from data and technology leaders sits in the timeline. Ofwat intends to consult with the sector in autumn 2026, ahead of a first iteration of AI guidance expected to cover principles, data readiness standards, and an industry taxonomy. Water companies are also expected to have responsible AI governance in place, with data readiness, transparency, accountability and security controls all named explicitly.
Read that list again….
- Data readiness
- Transparency
- Accountability
…and you’ll realise that none of it is AI-specific.
It’s data governance, given a new deadline because AI has made the cost of getting it wrong faster and harder to unwind.
Ungoverned data isn’t a new problem, but an old one with a deadline
Every water company already knows what happens when a number can’t be traced to its source: a regulator asks where it came from, and “we’re confident it’s right” isn’t an answer that survives scrutiny.
AI just raises the stakes.
A model built on ungoverned data doesn’t remove that risk, but produces confident answers at a speed that nobody can check, which is the last thing a regulator, a customer or a court wants to see.
That’s the uncomfortable part of Ofwat’s plan for companies that have treated data governance as a compliance chore rather than a foundation. You can’t bolt AI onto data nobody has cleaned for years and expect insight. You get outputs that look like insight, and a growing gap between what the model says and what the company can actually stand behind if asked.
Find more information on the risks of poorly managed and ungoverned data in our blog: Why invest in data quality?
Three questions water companies need to ask before Ofwat’s AI consultation
Rather than being the ones with the most AI pilots, the companies walking into autumn’s consultation from a position of strength will be those who can already answer these three questions:
1. Are your AI models trained on the same trusted data your regulatory reports use?
If the answer is “different systems, different pipelines,” that’s the gap Ofwat’s data readiness standards are designed to close.
One version of the truth has to exist before AI can be built on top of it.
2. Could you show, in a meeting, who has access to a given dataset?
Not in principle, but on the spot.
If governance lives in a policy document nobody has opened this year, this question exposes it fast.
3. Is governance enforced by the platform, or written somewhere nobody reads?
The difference decides whether governance survives contact with operations.
Rules that live in a document get worked around under pressure. Rules built into the platform don’t need anyone to remember them.
All figures and timelines drawn from Ofwat’s AI adoption plan for the water sector, published June 2026.
How to prep for Ofwat’s AI consultation
Answering “no” to any of these isn’t a reason to panic, but a reason to start now. That’s on one dataset or one process, rather than waiting for autumn’s guidance to arrive and trying to catch up all at once.
It’s also something Oakland Everything Data can help you with. Our experienced AI consultants work with our water sector specialists to bring your data and AI governance up to speed ahead of this autumn’s consultation. Our Artificial Intelligence Consulting Services page explains what we do, and why, or you can get in touch with us directly to see how we can help your company.
Governance that pays for itself
The instinct with data and AI governance work is to treat it as an overhead necessity rather than something that shows up on a business case. The evidence from inside the sector says otherwise.
In one governance programme we’ve delivered for a major UK utility, a single customer-data use case identified £85 million in value once governance was embedded into a core business function rather than run alongside it. That programme now runs on four data owners, five lead stewards and 28 trained stewards, with dashboards and a centralised platform making ownership and data issues visible to everyone who needs to see them, not just the data team.
That’s the shift Ofwat’s plan is pushing the whole sector toward, whether individual companies are ready for it or not.
The same pattern shows up in Open Data. Yorkshire Water needed a strategy for managing data shared externally and a process for deciding what to release. Oakland worked with them to define an operating model and governance framework, which moved them to the top of Stream’s publishing rankings. Instead of being the obstacle to openness, here governance was the thing that made it possible, all without the risk. Read our Yorkshire Water data strategy case study for all the details on how we brought this data strategy to life. You may also be interested in our blog: Why Open Data is the future of water regulation.
“Governance stopped being Oakland versus us and became one team working to the same agenda.”
Rob Lancashire, Data Management Lead
Where to start?
Ofwat’s data readiness standards won’t be published in full until after the autumn consultation, so there’s no finished checklist to work through yet. But the shape of the ask is already clear enough to act on: trusted data, traceable lineage, ownership that’s named rather than assumed, and governance enforced by the systems people actually use.
That’s not a five-year data strategy. It’s a starting point – one dataset, one process, or one use case, governed properly and proven fast, then built on. We believe the companies that treat this as the next compliance deadline will spend fourteen months preparing a response. For those treating it as the outcome they were already working toward, well they will spend fourteen months getting ahead of it.
For support getting your data in ship-shape to align with Ofwat’s latest plans, please contact our team.
