AssureAssist is a smart move. Here’s how to build on it
- by Craig Dixon, Solution Specialist
- 16 November 2026
- Approx 5 min. read
AssureAssist – the new AI tool from NISTA
Government has built something extremely useful - AssureAssist is an AI tool developed by the National Infrastructure and Service Transformation Authority (NISTA) with i.AI (the government's Incubator for Artificial Intelligence), which started life under the name Scout. It reads the documents a project submits for an assurance review and points reviewers towards the questions most worth asking. In testing, it halved the time review teams spent preparing, which matters a great deal for a process that has always leaned on a small pool of experienced people giving up their time.
I've spent years around programme delivery and assurance, first in local government and now as a Solutions Specialist at Verto, where much of my week is spent with PMO teams talking through how they really run their portfolios, programmes and projects. AssureAssist is a good idea, it has been built with care, and the people behind it deserve credit.
The point of this post is to highlight the benefits AssureAssist brings and explore how we can build on them. It’s about how complementary tools and support can help teams get more from the review process, while strengthening assurance throughout delivery.
What AssureAssist does well
Anyone who has been part of an assurance review knows how much work goes into the process. The team review wealth of information and spends hours reading business cases, plans, risk registers and more to pinpoint where the issues and non-compliance might be. It's important work, but a lot of it is reading and cross-referencing, and that's exactly where AI can help.
AssureAssist takes on that first pass. It checks the documents against NISTA's own review handbooks and gate criteria, flags what looks thin or missing, and links every finding back to the page it came from so the reviewer can check it for themselves. In beta testing it was 92.5% accurate at identifying the right lines of enquiry, which is a strong result for a first version of any tool.
What I like most is the thinking behind it, it's designed to support reviewers rather than replace them, so human judgement stays in charge of the conclusions. It's grounded in government's own guidance rather than a generic idea of what good looks like, which matters when the stakes are this high. It's ready for Government Major Projects Portfolio reviews now, and NISTA has said it will roll out across all projects and programmes soon.
What it isn’t designed for
The distinction is that AssureAssist reviews what you give it; it isn’t a live assistant you can call on throughout delivery to check quality and compliance as you go. You need to gather the information and manually feed it into the tool each time you want it to carry out a review. That makes it useful for reviewing a pack of evidence, but it isn’t the same as having ongoing assurance built into the way a project runs.
Each review is therefore a snapshot of the information supplied at that point. Once a change happens, that picture can move on. For AssureAssist to review the new position, someone must provide the updated information and run the review again. The gap isn’t just about what the documents contain, it’s about what happens between reviews.
The live picture still sits somewhere else. Risks are being managed regularly in spreadsheets, plans being tweaked and refined as things progress. The business case itself goes through many iterations too, from minor adjustments to major shifts in what the project is setting out to deliver. At portfolio level, it also means seeing where projects are competing for the same people, budgets and suppliers. This is where routine assurance support can make a difference: helping teams test their assumptions, check quality and compliance, and use current information to guide decisions as the work progresses. It’s about informing delivery while there’s still time to adjust, rather than waiting for the next formal review. That complements the point-in-time assurance AssureAssist provides, giving teams support between reviews as well as a clearer picture when a review comes round.
What local government taught me
When I worked in local government I was lucky enough to work with some of the most committed public servants I've known, holding initiatives together with little support. The challenges were rarely about people, they were about capacity, tools and timing. PMO teams were small and often stretched across far more projects than they had people for, information lived in spreadsheets, shared drives and inboxes, and pulling it together for a board meant a scramble in the week before the meeting.
The result was that reports could be out of date by the time they were read, even though everyone had worked hard on them. RAG ratings tended to lean towards optimism, as they do almost everywhere, because teams genuinely believed they could recover slipping initiatives. Boards made decisions in good faith on information that was a week or two out of date and I don't think any of that is unique to councils, it's what happens anywhere reports have to be assembled by hand.
Councils sit outside the central government portfolio and nothing published suggests AssureAssist is heading their way. But NISTA's own annual report says it wants good delivery to be the standard across the whole of government, not just the big central projects. The lesson travels well, whatever the size of the organisation, a review is only ever as good as the evidence underneath it, and evidence that has to be rebuilt by hand for every review will always struggle to keep current.
Where PPM comes in
Verto started life in a council, and today more than 100 public sector organisations use it to run their portfolios, programmes and projects. When a programme runs in Verto, most of the evidence an assessor wants already exists as live data, rather than as a document written for the occasion.
Risks, assumptions, issues and dependencies sit in one place, with owners and ratings that change as the work does. Milestones and plans are tracked as delivery happens, so slippage shows up when it occurs rather than at the next board meeting. Finance and benefits are followed against the business case all the way through delivery, not just at the start. Approvals and workflows are built in, with an audit trail of who changed what and when, while portfolio dashboards and resource management show where programmes are pulling on the same people at the same time.
Our Verto Intelligence Partner, 'Stevie Standards', is an AI Agent and builds on that live picture. It is designed to provide Programme and Project Data Standard guidance directly within Verto, looking across a single project, a programme or an entire portfolio. Alongside compliance analysis, it offers practical support and coaching, highlighting potential gaps and areas for improvement so teams can address them as delivery progresses. It’s a powerful way to make compliance part of everyday delivery, rather than a task left until review time. The aim is that, by the time documentation is uploaded to AssureAssist, teams have already identified and addressed gaps against the Data Standard, so the evidence should meet its requirements before the formal review begins.
That support complements AssureAssist rather than competing with it. Verto exports reports straight to Word, PDF and PowerPoint, so the review pack can draw on the same live data teams have been checking and improving, instead of being written from scratch the week before. Reviewers still get the documents they need, but those documents come from evidence maintained throughout delivery, not assembled just for the occasion.
Use AssureAssist, but build assurance into delivery
I hope AssureAssist is widely used. It offers valuable support to reviewers, and making formal assurance more efficient is worth doing. But I’d encourage teams to use it as part of their assurance approach, not rely on it as the whole answer. Keep the evidence live, use supporting tools and guided AI where they can help, and give people the information and support they need to address issues throughout delivery, not just when a review comes round.
The Programme and Project Data Standard, due to become mandatory from January 2027, makes that approach all the more relevant. A common structure for project information provides a foundation, but the real value comes from keeping that information current and using it to test assumptions, identify gaps and guide decisions. Compliance should be something teams work on as they deliver, rather than something they try to demonstrate at the end.
Government Project Delivery has said it wants assurance to become continuous rather than a point-in-time exercise. That’s an ambition I share, but supporting it doesn’t have to wait for the next generation of tools. Our Verto Intelligence Partners already bring guided AI support into the delivery environment. 'Stevie Standards', for example, helps teams assess alignment with the Data Standard, understand potential gaps and identify improvements across projects, programmes and portfolios. It supports the work between formal reviews, while leaving judgement and accountability with people.
My advice to SROs and PMO leads is simple: use AssureAssist, but don’t wait for it to tell you where the gaps are. Build assurance into the way you run your projects, so the review pack reflects evidence you’ve already been maintaining, checking and improving. That gives delivery teams support when they can act on it, and reviewers a stronger starting point when formal assurance begins. If you’d like to see how that works in practice, get in touch with the Verto team and ask for me.
Podcast
The AI Adoption Gap: What's holding public sector organisations back?