National Grid's Peter Hancock at nPlan Summer AI Day 2026: The Highlights

Just over a month ago we held our biggest AI Day to date, streamed live from Kachette in London, on the theme of Construction Superintelligence. If you read my last post, you'll know that we defied a red weather warning to keep our show on the road - and it turned out to be the right call, with our keynote--which featured the launch of Decision Intelligence, our Khaleesi orchestration engine, and a preview of Barry Live--getting a big (positive) response from the audience.

But the keynote is only ever half the AI Day story - as regular attendees will know, we have a track record of convincing brilliant folks who are actually out in the field - in leadership roles on megaprojects - to come and talk about their projects and their experiences using our AI tools. This Summer's event was no exception, with one of our fireside chatterers--Peter Hancock--joining to talk about his work as the Project Director of National Grid's Didcot Substation.

Pete is a pretty humble guy - he didn't mention that Didcot will be the UK's largest substation when it completes in 2029 - as well as a key piece of the UK's AI infrastructure. I guess I've just done it for him.

In this blog I've pulled together what I felt were the most interesting and intriguing Pete moments from the fireside chat - starting with what he did say about Didcot. Take a look👇

Keeping the lights on while you build

The complexities of pulling together all of the aspects, both for a construction project and an electricity transmission project, pulling it together so that it works, it meshes with a live network that is sitting adjacent to our, our project so, so keep the lights on while we're building to connect in and to power customers at the end all requires vast amounts of information to bring the confidence and the, the surety that we are on track, time and cost.

nPlan has long talked about the increasing complexity of large-scale projects - and the role that complexity plays in making them difficult--if not impossible--for humans to forecast and de-risk. The need to 'keep the lights on' while Didcot is upgraded and the project's exposure to the very real risk of disruption to international shipping routes only adds to the complexity (and the pressure Pete is under), which is likely part of the reason Pete turned to a machine (nPlan's AI) to help with project assurance.

Happily we don't have to guess - Dev's next move was to ask Pete why he'd decided to shake up the status quo and work with nPlan:

Why shake up the status quo?

Status quo works, status quo functions… I think the AI space, the big value add I see is it gives far more direct access to your schedule. So examples just up on the screen a few minutes ago, I could log on in the morning and start to ask, challenge, and probe areas. Start to ask questions around the schedule, and I get direct access to it as opposed to waiting for the regular monthly cycle and updates that way.

Pete's answer here was genuinely unexpected - AI adopters often talk about countering the malign effects of cognitive biases and taking a data-driven approach to assurance - but Pete brought up how AI has enabled him to go from monthly to daily assurance. The implication is that AI has enabled Pete to ditch a static and reactive approach to project assurance in favour of a dynamic and proactive one. It's what we expected to happen when we set out on the most recent stage of the nPlan journey, but it's still awesome to hear Pete confirming the hypothesis live on stage.

The aha moment: noticing the small activities that derail a schedule

One of the early things it did was highlight that some of our smaller activities, which were not getting lots of focus, had the potential to be quite disruptive to the, the plan schedule. So lots of small utility diversions, water, small power, sewage, those kinds of things that do not draw the instant attention that a factory acceptance test might for a large transformer in a international factory somewhere.

What Pete is describing here is a textbook case of salience bias - the tendency to fixate on the flashy, high-profile risks (the multi-million pound transformer being built in some far-off factory) while the boring stuff (a small utility diversion) quietly threatens to derail the whole schedule. It's a bias that AI is particularly well suited to counter - an algorithm doesn't care how sexy a risk is, only how likely it is to cause delay - and it's a benefit that nPlan customers bring up again and again. Anglian Water are a great example: the first schedules they ran through nPlan flagged emergency showers - not transformers, not pipelines, but showers - as a leading risk, having spotted that these tiny installations were forever being delivered late. So Pete is in good company.

Driving adoption by leading from the front

The big one for me was shifting to our using this for our monthly reporting information. So saying, "This is how I want the information. This is how we're gonna present it." I think that drove then the adoption and it, it brought the, the, the accountability and the seriousness to using it as well.

We've already heard Pete say that the real value of AI is in using it for daily assurance - the ability to log on any morning and interrogate the schedule - thereby freeing him from a reactive monthly reporting cycle. And yet that same cycle gave Pete the cover he needed to drive adoption. Why? Because monthly reporting was already part of the furniture - a workflow his team knew, trusted and did anyway. By making nPlan the way that familiar job got done, Pete gave people a low-stakes on-ramp to the new tech, sidestepping the defensive 'why are you changing how I work?' reaction that so often kills adoption stone dead.

Winning over the skeptics

Proving benefit, proving it might not always be realizing time improvements if we keep on the schedule, but demonstrating knowledge of risks, demonstrating increased probability of delivering, I think those things are, are valuable, and being able to articulate that benefit over a, a cost of some sort is the key item to do.

Notice how conservative Pete is here. He doesn't recommend trying to win skeptics over with grand promises about slashing months off the programme. Instead he makes the case that increased confidence in outcomes is already an example of AI delivering outsized value - and uses that as his starting point.

An independent source of assurance

It's the AI will ask lots of questions. It will… We, we provide it data, we provide it information, and it challenges, it probes. It doesn't bring assumptions. It doesn't assume that you've got a farm producing enough bentonite. It, says, "Have you?" And then you go, "Oh, I haven't." So it brings early visibility.

Here Pete mentions something which has always been one of nPlan's strongest selling points: AI as an independent source of assurance. Because it holds no assumptions and carries none of the 'we've always done it this way' baggage that humans inevitably bring, the system asks the awkward questions a project team might not think - or want - to ask itself. It's Pete showing his awareness (again) of AI's power to counter our cognitive biases - even though, as we saw right at the start, countering bias wasn't the reason he brought the tech in. He came for the daily access to the schedule and proactivity; the independent challenge is a further bonus.

Pete's advice: be patient

Be clear on problem statement that you're trying to tackle. Understand what quick wins you've then got in that space. So am I schedule sensitive? Do I want to understand greater information around risks to a delivery?… And then celebrate the small wins and try and cascade that into a bigger change piece.

Once again, Pete is conservative but sensible: bring people with you, celebrate the small wins, and don't try to rush it. Be patient.

Watch the full fireside

Those are my favourite Pete moments, but there's plenty more where they came from - including Sara Loureiro's view from BCG on the state of AI adoption across capital projects. You can watch the whole of AI Day right here.

In the meantime, watch this space for my next blog on Sara Loureiro of BCG's best fireside chat moments - it'll be live here soon.