Every publisher understands the pressure of a deadline. But most deadlines, when push comes to shove, are negotiable. Ours is not.
Hansard – the Official Report of Parliamentary Debates – must be published every single sitting day without exception. That obligation has been met continuously since 1909. It is protected under the Parliamentary Papers Act 1840. It is parliamentary convention, a constitutional necessity, and the official report of what was said in Parliament. There is no equivalent of ‘we’ll post it tomorrow’ or ‘the print run was delayed’. When the House rises, the clock starts. Digital publication must be live within three hours of proceedings concluding, and the print-ready PDF must be production-ready before 6am.
I manage a team of dedicated publishing professionals who orchestrate that cycle – night after night, session after session, Parliament after Parliament. The last thing any of us wants to do is change something that works. And yet, standing still is no longer an option.
115 years of no-fail publishing
It is worth pausing to appreciate what that publishing record actually represents, before getting to where we are headed.
On any given sitting day, Hansard reporters begin transcribing proceedings as soon as the days’ business is underway. XML files arrive into our production system throughout the day – hearings, debates, written statements – continuing until three or four hours after each House has finished. The production team then works through the night to validate, compile, format, index and publish. By the early hours of Tuesday to Friday mornings, the day’s record is complete. It then goes to print and to our online platform, hansard.parliament.uk.
Now multiply that across Commons, Lords and Parliamentary Committees – often running in parallel. Add the complexity of Westminster’s procedural conventions: the rigid sequential numbering of issues, columns and bound volumes, the precise formatting rules, the requirements for corrected and uncorrected versions. Then consider that this all happens under the additional constraint of parliamentary security, data sovereignty requirements, and the sheer constitutional weight of getting every word right.
This is not a publishing operation that lends itself to disruption for disruption’s sake.
Jack Homer, Hansard’s editor, puts it well: “The reason our workflows have been stable for so long is that they had to be. The Official Report is a legal record and a democratic one. Every word matters. Every publication matters. The integrity of the process is non-negotiable – and any evolution has to begin from that foundation, not in spite of it.”
The case for change
That said, the publishing technology landscape around us has changed beyond recognition since our current workflows were first designed. The platforms that publishers now take for granted – cloud-native architecture, API-driven content delivery, real-time digital publication, integrated validation, structured content that can be repurposed across channels – were either unavailable or prohibitively complex when Hansard’s current production model was first established.
Any publisher operating a legacy system will recognise the tension. The older a workflow becomes, the more it works – in the sense that everyone knows it, has worked around its quirks and built their day around its rhythms. But the older it becomes, the more it also calculates against the future: harder to extend, harder to integrate, and increasingly reliant on tribal knowledge rather than robust, documented process. Legacy is the price of reliability, paid in arrears.
For Hansard, modernisation isn’t about chasing trends. It is about building the right foundations to ensure that the Official Report remains as fit for purpose in the next 115 years as it has been in the last. That work is now actively underway.
A new foundation for production
The most immediate pillar of change is a project to significantly upgrade and enhance our content management system with a modern, cloud-deployable, XML-native platform. Our strategic intent is clear: to move our production workflow from a system that has served us well but reached the limits of its adaptability, to one designed for the realities of contemporary publishing infrastructure.
The requirements we set out for any replacement are revealing in themselves. We need native XML database architecture rather than traditional SQL storage – because every item of content we produce is XML, and our system should treat it as such. We need integrated validation built into the editorial interface rather than handled by a separate desktop application. We need version control, peer review workflows, configurable automated processing and a modern browser-based user experience that reduces training time and eliminates manual workarounds.
Critically, we need a platform with the capacity to expand. Our ambitions extend beyond the Hansard workflow to the full range of the Procedural Publishing Unit’s output, including the business papers that form the daily scaffolding of parliamentary life: the Order Paper, Amendment documents and more.
This is a significant undertaking, and it is being approached with the same care we bring to every publication: methodically, with extensive testing, and with a parallel-running period to ensure zero disruption to the overnight production cycle. If that sounds conservative, it is meant to. The record that has been maintained continuously since 1909 will not be put at risk in the pursuit of innovation.
Hansard online: from archive to engagement
Beyond production, the digital face of Hansard has also been evolving. Hansard Online – hansard.parliament.uk – has come a long way since Hansard first appeared on the internet in 1997. The platform was significantly overhauled in 2018, introducing mobile responsiveness, improved search, individual Member pages with permanent URLs, and enhanced division data including downloadable results and visual breakdowns by party. We also recently completed a long-running project to include content on the site that had previously been available only as paper bound volumes; earlier this year we celebrated the fact that the complete Hansard archive going back to 1909, and including reports dating back to 1803, is available online.
But the opportunity ahead is considerably larger. The content we publish is some of the richest structured data in UK public life. Every speech is attributed to a named Member, tagged with date, constituency, party and role. Every vote is recorded. Every written statement, every committee hearing, every question and answer – all of it, going back more than 200 years in digitised form.
The question is: how do we make that data more genuinely useful, not just more accessible?
There is growing demand from researchers, journalists, civil society organisations and civic technologists to work with Hansard data programmatically – to build tools, conduct longitudinal analysis, explore how debates have shifted across generations, hold representatives to account with evidence. The Parliamentary Digital Service has been developing API infrastructure to support this, and it is an area we expect to see develop significantly in the coming years, aligned with Parliament’s broader 2026-2030 Information and Digital Strategy.
The AI horizon
The most consequential long-term development, though, is artificial intelligence – and here the conversation in parliamentary publishing circles is both more exciting and more careful than in much of the broader media industry.
Hansard is, in structural terms, exactly the kind of content that AI handles well: highly formatted, consistently attributed, rich in metadata, with patterns and conventions that repeat reliably across decades. The opportunities are considerable. AI-assisted semantic tagging, automatically linking mentions of legislation or Members to the appropriate data objects, could dramatically enrich the searchability of the record. AI-driven summarisation tools could make the daily output of both Houses more navigable for citizens without a law degree or parliamentary encyclopaedia to hand. AI transcription and quality-checking tools are already transforming how some parliaments around the world handle the production of their own official records – a trend tracked closely by the Inter-Parliamentary Union.
Internationally, parliaments are at very different stages of this journey. The IPU has published guidance for smaller legislatures on AI transcription as an entry point. Canada has created a dedicated Ministry of Artificial Intelligence and Digital Innovation. Several Caribbean parliaments have built open-source, API-first platforms with AI-integrated records management from the ground up, unconstrained by legacy systems.
For the UK Parliament, the path is more carefully managed – as it should be. Hansard is not just content; it is the official report of proceedings and carries constitutional weight that demands the highest standards of accuracy, attributability and integrity. Any AI application touching the production or delivery of that record will need to meet a correspondingly high bar. With that in mind, we are proceeding with careful experimentation and trialling of AI tools. Commons Hansard is already using speech recognition software to support Hansard reporters in their job, and we also have a live beta project that is helping with transcription.
Jack Homer is clear-eyed about both the opportunity and the responsibility: “Hansard must be authoritative and we are treating capabilities such as summarisation or sentiment analysis with caution. At the same time, we see great potential in semantic search and intelligent editing of text produced via speech recognition. We are not going to be left behind, but we will not introduce AI into the production chain where it is not appropriate or for the sake of doing so. The question we ask is: does this make us more efficient, and does it make the record more reliable, more accessible, or more valuable – without introducing new risks to its integrity? If the answer is yes, we move forward. If not, we wait.”
That is a reasonable standard for any publisher to hold. But the direction of travel is unambiguous. A more modern production infrastructure – which we are now building – will make it significantly easier to deploy AI tools responsibly in the future, because the content will be cleaner, better structured, and governed by more robust, documented workflows.
What this means for publishing more broadly
I recognise that most InPublishing readers do not run parliamentary publishing operations. But I suspect the underlying dynamic is familiar: a heritage publication with a loyal audience with high expectations, workflow habits built over decades, a legacy technology stack that is both the reason you’ve maintained reliability and the reason you’re struggling to move forward, and a growing imperative to make your content work harder across more channels.
The lesson from Hansard’s current transformation is not that caution is old-fashioned. It is that caution and ambition are not opposites. The right response to a 115-year publishing record is not to bet it on a single disruptive leap. It is to build the foundation carefully, modernise incrementally, and ensure that every improvement serves the core mission rather than complicating it.
The Official Report of Parliamentary Debates will be published tomorrow. And the day after. And the day after that – whatever the technology behind it, whatever the tools that help produce it, whatever the channels through which it reaches an ever-broader audience. That is the commitment. The transformation is simply about making sure we can keep it – and do far more with it – in the coming months and for the next century and beyond.
This article was first published in InPublishing magazine. If you would like to be added to the free mailing list to receive the magazine, please register here.
