How the rise in the use of AI chatbots is encouraging companies and design agencies to ditch PDF-first reporting processes.

Introduction

Companies invest a huge amount of time, resources and money producing detailed and trusted annual reports each year, but few people read these reports in full. Most reports are used as a vital reference point for specific content about any topic a user may be researching.

In an AI world, the annual report has become more important than ever. It is approved by the board, audited and contains the deep information which investors and other stakeholders need, and which can’t be found in complete form anywhere else.

With AI searches now dominating most users’ access to company information, annual reports need to be published in a format where the content of the report is AI ready. But unfortunately, in most reports this crucial data is still locked inside PDFs which cannot be accurately searched by AI tools.

Regulators have introduced modern digital formats, based on HTML and enhanced with digital tagging (known as Inline XBRL), to help them supervise companies and to encourage capital market assurance. These new formats make annual reports easily readable by both humans and computers.

Today, most companies in Europe still begin the process of building an annual report by generating a PDF document and then converting it to iXBRL after sign-off. Unfortunately, in doing this, many of the PDF format’s underlying structural flaws and errors are being carried into the digital version.

So, why is annual reporting still so reliant on print-first PDF software and why is it taking so long to digitise the processes?

This article looks at the history, how the move to AI will impact annual report production, and the significant role of design agencies in digitisation over the coming years.

Annual Reports Remain Limited by PDF-First Software

From an investor point of view, company annual reports are the most vital information that companies provide as they contain much of the information you need to assess a company — management statements on its strategy and objectives, updates on operations, risk and governance, and detailed financial information accompanied by an auditors statement.

The annual report also provides similar datasets across all companies using reasonably consistent standards, and this also plays out internationally. This helps to make a wide mix of reporting content and data more comparable, so it meets many different needs for diverse audiences — including investors, customers, employees, regulators, recruits, and partners.

Corporate reports started to move from physical print focus to PDF in the mid-90s and had become largely PDF-first by the mid-2000s.

But although PDF is an electronic format, it is not really “digital”, which means the information it contains is not properly readable by machines. Over the past 15 years, efforts to make PDF appear to be more “digital” has resulted in many bolt-on tools, but at their core these are all still PDF-first formats, and not accurately readable by search engines.

The PDF format is often referred to as “paper under glass’ — the information published in PDF files is unstructured, inaccessible, and hard for computer systems to access, read and understand. They are read like a printed book with no dynamic links, as is the case with a website, and do not automatically reflow text or resize images to fit the screen size. As a result, PDF files fail to fully meet accessibility standards and cannot be read on mobile devices.

Regulators Move to Modern Reporting Standards

Over the past 15 years major regulators around the world have recognised the limitations of PDF reports and have moved their reporting frameworks for annual reports to Inline XBRL. This is simply an HTML (web) format with embedded tags.

• The HTML format enables humans to find, access and read the content in any browser or device. i.e. just like any good website the report can respond to the device. When they are well structured, search engines can interpret their layout.
• iXBRL tags give precise meaning to every data point. e.g. the tag tells you that this is ‘revenue’, for ‘company A’, for ‘period x,’ and in ‘unit y’. This enables the data to be read, compared and analysed accurately.

There are differences between regional reporting frameworks that affect both the structure of financial reports and how XBRL documents are produced:

• US SEC (US GAAP): Filings are extensive and detailed, but presentation generally follows well-established filing conventions and structures, with a relatively consistent layout across issuers.
• Europe (ESEF / UKSEF): The annual report is a broader communication document, designed for a wide stakeholder audience. Reports are presentation-driven, highly formatted, and often strongly branded.

These differences create challenges for companies reporting across jurisdictions and require software vendors to support multiple production approaches, despite using the same underlying global XBRL standard. It is hoped that further standardisation of global reporting practices will iron out these differences in the longer-term,

The European model introduces additional complexity in aligning visual presentation with the XBRL tagged data. This article focuses on the European model and ESEF, while noting that many observations are relevant more broadly.

Despite the regulatory push to digital reporting, five years into the ESEF mandate most European companies still rely on a PDF-first approach, digital-second process. i.e. producing a PDF report and then converting it to iXBRL. A process that carries over many of PDF’s flaws:

• Conversion of PDFs absolute content placement approach to HTML delivers poor document structure, large file sizes, and poor usability.
• Even the converted HTML output is not compliant with global accessibility standards (WCAG 1.4.10).
• There is an increased risk of tagging errors as XBRL tagging only occurs after the PDF report has been finalised, at the end of the process. Sometimes it is too late to fix errors introduced by tagging unstructured content.

The PDF Conversion Paradox

The fact that most companies still use a ‘PDF-first approach, digital-second’ approach shows that there is a paradox at the heart of the current annual report frameworks.

The differing goals of stakeholders are compromised by the PDF-first process and are enhanced by the digital-first process.

• Regulators want to be able to scrutinise volumes of reports faster and more reliably and reduce errors in the historical data record. However, the European Securities and Markets Authority (ESMA) which sets the technical reporting standards for ESEF is not yet ready to enforce the quality of the HTML in its Filing Manual, while most local national authorities that collect the reports do not have any sanctions for tagging errors, possibly worried by the negative feedback from companies forced to change a well-established process.
• Investment managers need company data based upon trustworthy sources that they can readily review and compare easily. Despite widespread adoption of iXBRL for data analysis in the US, there is no loud clamour from European investment firms to use the published XBRL data. Why is unclear, but it may just take time to adopt new technology platforms.
• Auditors would like verification processes that enable greater accuracy and mitigate the risk of errors. However, I have heard many auditors complain that the XBRL tagging requires additional resources and increases staff costs, rather than automating the process.

Even those companies trying to produce compliant, error-free reports and claiming that they are fully committed to meeting accessibility standards, often only provide the PDF download of the annual report on their website and still don’t make a simple readable version (iXBRL viewer) of the mandatory digital report available.

So, despite its many flaws, the PDF-first process continues to be the default approach for most companies. Unfortunately, this leaves a poor historical record for those wishing to analyse the data in future.

The Alternative: Digital First

Digital-first means literally creating reports natively in HTML from the outset. A high quality digital-first report has best practice structure (as with websites), the output is responsive on mobile and can fully meet accessibility requirements. The digital-first process also provides the platform for producing a PDF version which will continue to be the format of choice for some users.

There are several software tools available that offer a digital-first approach, and I reviewed these in a previous article — ‘Digital First is Here’. These digital design tools are often based on a web content management system (CMS) that enables multi-format output (HTML, PDF, iXBRL) from a single digital source of truth.

I found in researching these tools that the support for XBRL tagging is basic, and most are yet to be proven in supporting complex processes. However, the inline XBRL production process, and particularly the visually rich, European ESEF report, is much better supported by these software tools.

There are some good examples of digital-first reports already available:

• Airtel Africa Annual report 2025
• Arcadis Annual Report 2025
• Irish Continental Group Annual report 2025

Through integrated native HTML workflows, the digital-first approach can significantly improve the quality of the data, as it enables tagging throughout the process so that data errors can be found and fixed during the report creation process, not at the end.

Given these rapid advances in native HTML design tools which deliver higher quality, digital-first reports, it does surprise me that only a few innovative companies have so far moved from PDF-first to digital-first processes.

Especially when, research of user analytics data from early digital-first reporting by Friend Studio has shown that:

• In the first year of providing a digital report, 14 times more visitors were attracted, and over 2 years (across 5 reports) 60 times more visitors were attracted to read the online report than the prior years’ downloaded PDFs.
• Digitisation also resulted in a 10x increase in total PDF downloads versus the PDF-only reporting in the prior year. This was a result of greater search engine visibility
• The online version also allows analysis of what users were looking at, which the PDF version could never do.

The Crucial Role of Design Agencies

Last year (2025), I decided to dig deeper and spoke to a number of design agencies across Europe, to understand why companies were still using the PDF conversion approach. I also wanted to assess if it was in part the design agencies themselves which were holding back the move to digital first, i.e., were there misconceptions about digital reporting, fear of change, protecting their existing business models, dependency on Adobe InDesign, costs and time of training staff in new digital processes, etc.

The first piece of feedback that all the design agencies that I spoke to made was that it is the reporting company that selects the approach to creating the annual report, i.e. most companies decide to use the PDF conversion approach due to the familiarity, years of developing processes for approval, and a cautious approach to what is a very complex, multi-department process.

Some design agencies are now offering their clients multiple approaches, including digital-first reporting options. But as I researched, I found very few design teams which were experts in the wider issues discussed above. In fact, higher quality data that XBRL tagging delivers, and the benefits to investors using AI tools was hardly mentioned at all (… which I guess is not surprising given their focus is on design).

When quizzed, most design teams understand that HTML-first is the right way to move for accessibility and future proofing. But they are still cautious about moving to solutions that their clients don’t yet fully understand, and not yet confident about their crucial role in solving the digital transition for their clients.

What I did find surprising was the extent to which design agencies are at the heart of the annual report process itself. I learned that the design process usually binds together all the numerous departments involved within a company, and also the auditors and external accountants, investor relations advisers, and the XBRL tagging experts.

It is often the design agency which advises the internal report team and acts as the ‘glue’ combining the various internal and external contributors to produce the final report formats for approval, publication and filing.

Many agencies pointed to a half-way, hybrid approach — HTML summary page + PDF downloads. This is a popular idea, but is still “PDF-first, digital-second” — it does not future proof the data or solve the accessibility challenges that companies face in the years ahead.

So, most design agencies are aware of, but do not yet appear to be providing digital first solutions to their clients, which remain driven by the trusted approach they used for printed reports. Many companies may also have thought ESEF was only a compliance exercise for the finance team and not have realised that full digital reporting and accessibility are all solved by a digital-first ESEF process.

A comfortable inertia had settled in, that was, until AI chatbots arrived and investors started using them to analyse company information.

The Catalyst for Change: AI Readability

A Bank of England survey found that, by 2024, 75% of finance specialists in the UK used AI to research companies and an Investing.com survey in the US found that 65% of retail investors use AI tools to help inform investment decisions.

Academic, commercial and technical research has also comprehensively answered the question “Does AI need structured data?”. There is now extensive evidence to support the role that XBRL tagging plays in reducing hallucinations and improving AI accuracy.

I still chuckle at something I read from an accounting symposium that ‘AI means that there is no need for structure in documents and hence for XBRL’. Hopefully, mistaken assertions like this will fade with evidence, experience and education. This recent article is the latest to explain why: “XBRL Cuts AI Errors in Reading Company Filings, Study Finds”.

AI’s demand for accessible and well-structured data will increasingly put companies under pressure to deliver AI-ready, digital-first annual reports and provide data-rich, engaging online reporting.

According to design agencies, their clients have now started to raise such questions. They are asking how they can improve the chances that AI chatbots will find their annual reports, and how this critical information can become a primary source for more accurate AI analysis.

The answer to these questions is simple: move to a digital-first design process and publish the report as a fully tagged Inline XBRL website.

So, will this AI demand now convince companies to shift to digital reporting processes?

I believe so, as AI becomes the dominant approach to search and summarise data, companies will want their important information to be presented in the best way. Gaining trust and experience in new approaches and software platforms will of course take time, but the rising number of digital-first, AI-friendly annual reports now available shows that this shift has already begun.

I suspect that integration with accounting and office environments will be the next key feature for digital first systems. When design, tagging, validation and review occur in disconnected environments, it becomes harder to maintain consistency, decision-making, and demonstrate control.

By contrast, a more integrated process helps organisations retain institutional knowledge, support repeatable controls and governance, and reduce the risk that critical context is not lost between reporting cycles.

Conclusion

The fast growth in AI use means companies are suddenly incentivised to improve both human accessibility and the AI-readability of their annual reports. I believe this will inevitably lead to digital-first becoming the dominant design process, with PDF becoming a secondary output from multi-channel publishing systems.

Effectively, AI is redefining what “good reporting” looks like.

• For now, the ESEF model appears to be the testing ground for digital-first software given its visual presentation requirements, combined with XBRL requirements.
• I expect the new ESG reporting frameworks (ISSB, CSRD) will also quickly adopt digital-first practices and systems. These requirements have no legacy print systems to hold them back, and are narrative led, which is far more effective digitally.

As this landscape evolves, the result should be improved accessibility and data quality for all. Delivering on one of the key hopes for inline XBRL, as well as its embedded automatic validation, it creates a “flywheel” effect where the information contained in reports is ‘crowd checked’. Such a reporting approach delivers a continuous improvement cycle and supports the shift to ever more standardised, comparable reporting.

I believe that design agencies have a critical role to play in this. There are many related business streams that are likely to develop for these agencies, extending their services from communications and compliance into digital expertise across annual and ESG reports (just as they do in websites). I also expect the more technically focused agencies will add XBRL advisory and tagging services too.

The standardisation in software and production that this brings is also likely to appeal to larger enterprise software firms, which will look to include such features in their integrated Disclosure Management systems — effectively joining all the dots, from data sources directly to digital-first publication online.

The audit and accounting firms would also benefit from these more standardised processes and the ability to scale, as the annual report process becomes ever more automated and linked to existing GRC systems.

The shift to digital-first is not merely a technical or online reporting upgrade, it is a foundational change that brings a more transparent, data-driven era of corporate accountability, and those who lead this transition will help shape the future of how company reports are created, communicated and consumed.

For those wishing to follow the ongoing research into structured data, AI and digital reporting standards, XBRL International’s research publications at xbrl.org/news provide a valuable and regularly updated resource.

The author is Martin DeVille of AM2 Limited

This article was originally published on Medium.com as part of the Digital Reporting Made Simple publication.