Why the AI revolution in capital markets is quieter than you think

For the past two years, the AI discourse has become increasingly pervasive in all walks of life, and for capital markets it is no different. It has been variously hailed as a digital crystal ball, a replacement for the junior analyst, and a fundamental threat to the origination of the corporate narrative.  

Some have posited that the old ways of our approach and understanding of public markets will soon become outdated. We have seen dramatic sell-offs in software stocks amid fears that AI could render swathes of the industry redundant. Block Inc’s ruthless 4,000 person cut to their workforce because of gains to AI productivity sparked controversy recently, but not among investors who drove a 16% increase in share price on the day of the announcement.  

But as the dust begins to settle on the initial hype, a more nuanced reality is emerging across trading floors, research desks, and boardrooms. 

The revolution is indeed happening – but it is much more of a silent one than anticipated. Equitory’s recent survey of a broad set of buy- and sell-side firms reveals a market in the midst of a massive technical pivot, with AI usage expected to nearly double in the next 12 months. Yet, the most striking takeaway isn’t how much has changed, but how much has stayed exactly the same. The tools are new, but the hunger for clarity, the scepticism of noise, and the reliance on human judgment remain the bedrock of the capital markets. 

Clearing the Desk 

Walk into a major buy-side firm today, and you won’t find a sentient robot picking stocks. Instead, you’ll find a Portfolio Manager using Microsoft Copilot, Claude, or AlphaSense to do something remarkably traditional: reading. 

The buy-side is currently leading the charge in capital markets preparedness. A significant 73.3% of buy-side firms surveyed have already provided formal AI training to their staff, creating a stark divide with the sell-side, where only 33.3% have received similar guidance. But they aren’t using this training to reinvent the fundamental laws of finance. Instead, they are using it as a high-speed filter to boost their own productivity.  

The primary application for the buy-side is the summarisation of company materials—annual reports, RNS releases, and earnings call transcripts. What we are seeing is a simplification of process and a decreased burden, leading to greater focus on alpha identification. Plc annual reports and accounts run hundreds of pages but are rarely read in a linear fashion. We have long known that buy-siders do not have time to scrawl through pages and pages on company websites, there are simply too many companies in their universe. This is why we always tell our clients about the three-click rule, which is that you need to be able to find anything you are looking for on an IR site within three clicks, or investors move on.  

AI isn’t re-inventing the wheel, it has become the ultimate administrative assistant, sifting through the haystack to find what actually moves the needle. 

The Trust Gap 

Despite these efficiency gains, reliability will long remain a sticking point. While the adoption curve is steep, the trust curve is surprisingly flat. On a scale of one to five, buy-side investors rate the reliability of AI outputs at a modest 2.67, while the sell-side is even more sceptical at 2.22. 

The reason for this scepticism is what we call the Trust Gap. While tools like Claude and other GPT-wrappers are increasingly favoured for specific financial tasks, they often struggle with the subtle nuances of local markets – the UK context being a prime example of where generic models can trip. The survey feedback was blunt: AI can be too clever for its own good. Respondents noted instances where a model might estimate a 2025 data point that doesn’t exist, forcing analysts to watch it like a hawk to avoid a self-perpetuating loop of false information. 

For Execs and IR teams, this creates a new risk. If your disclosure is misleading or inconsistent, or buried in jargon, the AI won’t just miss the point, it might invent a new one to fill the void. In the past, bad disclosure led to confusion; today, it leads to a hallucinated narrative that can be distributed to thousands of terminals in seconds. Focus on clear, consistent narrative. How has the sentiment of what you’re releasing to the market changed period-on-period, and is this an appropriate reflection of your equity story’s journey over the past six months?  

Harnessing Benchmarking 

Where the technology is really adding benefit is peer group analysis and gives users a competitive edge. The ability to instantly benchmark a company against a global cohort – analysing not just financial KPIs, but disclosure styles, ESG metrics, and management tone – is one of the strongest use cases for analysts either side of the wall. 

We are also seeing the emergence of customisable digital dashboards where agentic AI can identify key IR actions in the background, or even determine defensive RAG (red/amber/green) statuses for current holders, in terms of likelihood of selling off. By harnessing massive proprietary datasets and broker notes, investors are getting a clearer, real-time view of who is at risk of leaving their register and where the next big targeting opportunity lies. Peer-group analysis and investor targeting are moving from quarterly exercises to automated daily functions. 

Back to Basics 

Ultimately, the ground is shifting beneath your feet less than you might think. The AI-friendly IR strategy isn’t about gaming an algorithm or using the right keywords; it’s about a return to the basics with an increased focus on clear and consistent outputs. 

1. AI is now the primary reader: AI tools prioritise structured data and clear, concise language. If your reports are overly lengthy or lack clear tables, the AI summary – the version your investors actually read – will lose the nuance you spent weeks crafting. 
2. The Rise of defensive IR: The most sophisticated IR teams are no longer waiting for sell-side notes to drop. They are running their own statements through AI-trained models themselves, to see what they pick up.  
3. The Human Premium: Perhaps somewhat surprisingly, as AI increasingly handles more and more of the day-to-day, the value of direct management access is skyrocketing. 33.3% of buy-side investors say they now rely more on direct Q&A and meetings to sense-check and contextualise what the AI is telling them. 

We are currently witnessing flux, where the speed of AI adoption is forced to compete with a persistent lack of trust in its accuracy. We don’t believe this is the nuclear moment oft-posited, but merely the next evolution of the investment process. In this environment, the most effective IR strategies are those that master the basics: transparent disclosure and direct management engagement, ensuring the equity story remains intact, no matter which lens the investor chooses to look through. 

The machine can process the data, but it still takes a human to define the value. For now, at least. 

Equitory recently completed a comprehensive, proprietary study on AI behaviours across the buy- and sell-side. We have produced a detailed analysis of the survey results, including practical implications for Boards and IR teams. The full report is being shared with listed and private companies. If you would like a copy, please contact us here.