This is the first in a series of articles on AI and its impact for info services, by role types, platform, industry, asset class.

The hype around AI can be breathless, though as you look at what is already possible with the technology, it is hard not to [already] be amazed. This post focuses on the technological, workflow (meaning end user financial services) and vendor opportunity, I’ll restrict my misgivings on market sentiment, AI bubble debates and exuberance to another post or, more likely, a different forum.

Positioning decision support tools and research

Before I dive into research workflows, I think it’s important to position them within the broader context and ecosystem of market data platforms, which support many workflows. A market data system is always a decision support tool, but a decision support tool is not necessarily a market data platform.

Returning to user workflows, obviously these platforms serve many user types, covering many asset classes and many workflows. A simple classification system for the range of user workflows might be:

The research, analysis and valuation pieces seem particularly well suited to AI support, given that research requires a huge amount of in-depth reading, summarization, interpretation and contextualization, model building in excel, etc. This research workflow is conducted by many personas: sellside research, buyside research, IB (DCM, ECM, M&A), IR or business strategy and, lastly, there are also numerous use cases within academia.

Market monitoring: slow your roll

At the other end of the spectrum is market monitoring. If you’ve ever walked onto a trading floor you’ll know that market monitoring is a whole collection of workflows, monitoring movements in every conceivable type of cash and derivative instrument, benchmarks, curves, spreads, reading newswires, transcripts, filings, research, tracking technical indicators, economic and other macro announcements, regulatory, political, policy and legal decisions, blogs, the list goes on. There are similar workflows, if perhaps less frenetic, for wealth management and for corporates.

The big market monitoring franchises like Bloomberg, Factset, S&P and LSEG will figure out where Generative and Agentic AI add value, accelerating a users ability to discern critical information, interpret and react to it – but I don’t see news windows, tabular data, charts of all varieties, all the flashing lights disappearing anytime soon. How agentic AI will be able to add value for these tools, what can be automated or augmented, that’s the fascinating journey we get to watch over the next half decade. Will an agent replace the monitor themselves? I doubt it.

AI in the research process

The opportunity for AI here is difficult to overstate. Numerous workflows could benefit from automation or augmentation:

Examples are already here: Citadel

There was an early December article as Citadel rolled out a new AI capability. Here is the Reuters article quoting Umesh Subramanian, Citadel’s CTO, as saying:

The tool can highlight risks and generate customized research and reading lists based on an investor’s portfolio, thereby producing relevant information and content for investors in a fraction of the time it would take a human researcher.

This is certainly the near term direction, over the next year or two. While this example focuses on the content consumption to summarization elements, I’m equally bullish on the M365 elements. Analysts spend their days in Excel and then, depending on the role, may then need to spend hours in PowerPoint as well. How a copilot or other AI delivers automation and time saving in these contexts is just as exciting.

---

No early retirement for the analysts (hold the cocktails)

I do not think AI is coming for the research analyst’s role anytime soon. It’s impact on the role, nonetheless, will be almost as profound. In the same way a pocket calculator removed hours of slide rule work, or word processors destroyed the typewriter, AI will reduce the amount of reading, summarization, drafting required.

Job losses will come. I suspect in the same areas they’ve always come. A reasonable if crude proxy for vulnerability might be if a function has frequently been offshored in the last two decades, it is probably a candidate for AI automation, or at least significant augmentation with fewer headcount. Think of help desks to chatbots, coding, KPO functions, etc. I suspect automation comes last to front office, as much due to regulatory requirements and management’s view of operational risk / liability management as to any natural moat.

Questions remain on accuracy and the risks of hallucination but ultimately I suspect that guardrails will ensure these tools are not given the ability to provide false data. Even with these challenges, it’s already exciting to see how a key financial markets workflow is evolving, while keeping the human in the driver’s seat.

AI simply becomes the new baseline, with an analyst, junior banker, IRO and so on now arriving in the morning to receive complete summaries of overnight research, macro news, etc. Valuation models will be ready, first drafts of their research report. What we are seeing is a step change is the starting point. We once had to manually calculate bond prices, build DCF models for each firm covered. AI isn’t replacing the role of the analyst, but it’s giving them a new tool that will supercharge their productivity. The difference between the analyst with these tools for automation or augmentation, and those without will be profound.

So while I can see some reason for nervousness, ultimately the power of these tools will give the human a better, faster view of new information, automation of the processes they then went through to turn that view into conviction, and thus more time to weigh up the implications.

For 20-30 years, the challenge for market participants in at least large cap public markets has been solving the signal/noise challenge. It was impossible to read every news story, transcript, filing, disclosure, regulatory report, research. To track every macro variable. What AI seems most able to do is solve that challenge, bringing summaries where once one needed to read a 120 page annual report, sustainability disclosures, or a transcript that came in during the night.

Hopefully, this synthesizing of data means that a critical nugget of information may not be overlooked, and the analyst reaches an inclusion based on an amazing view of the critical information. It’s all amazingly exciting, of course there are many steps to reaching that point. Over time, that’s where I’m most confience that AI will bring true transformational value to the research landscape.

---

---