When intelligence is artificial: focusing on the basics

AI Blog Banner

Last week we sponsored the Institutional Investors awards in NYC. It was a great event, and we were excited to be part of supporting some of the best research analysts in the world (many of whom are Street Context users!).

As part of that partnership, I had the opportunity to write a thought piece for their event magazine. I knew the majority of the contributors would write something about the current narrative (Generative AI) so I decided to go a little against the grain, and talk about all the much more basic and pressing concerns this industry faces. Specifically, while everyone worries about “artificial intelligence”, the real challenge is that most of what the sell side knows about its clients – most client ‘intelligence’ – is in of itself wrong, thus ‘artificial’. This can be due to incorrect data, improperly entered data, or lost data. We have an issue with perceived intelligence being artificial. Like all systems, AI is garbage in, garbage out. You can’t build anything until you have the right foundation in place. Here is what I shared:


As I was pondering what to write for this piece, I spent some time thinking about what would be covered by others. I have no doubt the soup du jour of ‘Generative AI’ will be well covered by many. They will talk about the promises of more efficient solutions with elegant approaches to solve problems you didn’t even know you had. They will talk about research reports being automatically written, summarization technology, models that automatically update themselves, and research distribution systems that make recommendations as if by magic. These are exciting ideas, but unfortunately, they are like gleaming towers being built on a weak foundation: without a solid base, anything else will ultimately be unstable, and lead to ruin.

Instead, I wanted to focus on something different: that this industry has so many foundational problems which make the basic quantitative understanding we have about our clients – intelligence – likely incorrect, never mind generative. Before we can build, we must fix that foundation. In this industry, most intelligence about clients is itself artificial, and before we build for the future, we must solve the problems of today (or in many cases, the problems of ten to fifteen years ago). If the intelligence, data, and insights about clients is incorrect, then anything done with it will likely be just as incorrect (or worse),  AI or no AI.

What do I mean? When I talk to our clients and other industry participants, it doesn’t take long for them to point out all the challenges they are experiencing in their current understanding of clients. The contacts in their research distribution system are woefully outdated and incorrect (with some contacts having a half dozen entries spanning twenty years of career moves). The interaction data in their CRM is a hodgepodge of different structures depending on who wrote it, and how much time they had. In our case, we hear that their email engagement data is inflated and inaccurate. In fact, we recently ran a study which pointed out that 60% of the ‘raw’ readership data we capture is generated by machines (usually anti-virus software), leading to a massive misrepresentation of client interest and interaction if not properly filtered. No one should have a 100% click rate (because as interesting as it might be, I can almost guarantee you that no client is reading your disclaimer link). The point I am making is that the basic client intelligence you think you have is likely wildly inaccurate. Before you think about your AI roadmap for the year, there is a lot more foundational (and boring) work that can be done to make sure whatever you end up eventually building can actually stand long term.

That’s what we’re focused on for the year ahead at Street Context: continuing to improve the fidelity, reliability, and accuracy of the data we provide our clients with, so that they can confidently use that data to drive their business forward. To do that, we’re continuing to update the infrastructure this industry runs on: email. We’re improving how we update, curate, and clean up contacts. We’re continually investing in our email infrastructure that delivers the fastest send speed in the industry, with the most accurate data (we delivered billions of emails last year, and that’s only growing). We’re focusing on growing the number of accurate data points we deliver to our clients (our largest client consumed over ten billion email interactions last year).

Most importantly, we’re also opening that ecosystem up to the buy side this year. We’re working with broker relations teams to more effectively distribute change requests including onboarding new investment professionals (and offboarding), sharing coverage changes, and any other information or updates. That is leading us to rethink subscription and preference management for our sell side clients as well.

There are basic challenges our industry faces when it comes to the flow of information: who wants what, who gets what, what they do with it, and what else they want moving forward. Truly solving those foundational problems is the critical path to being able to incorporate more exciting (and dare I say, intelligent) tools moving forward. For now, we’re focused on building the infrastructure that will deliver reliable and accurate client intelligence to our partners. The industry is well overdue for such a change. 

And rest assured, this thought piece wasn’t written by ChatGPT. We’ll save that for the future.



To subscribe to Blair’s thought pieces, click here.

My recent webinar: The Evolving Role of Salespeople in the Capital Markets

Previous Thought Pieces:

Bloomberg opens its ecosystem: chatbots to do your bidding

The future of salespeople: extinction or evolution?

When does an industry practice become “table stakes”?

Client coverage and relationship management in a post-COVID world

The end of MiFID driven unbundling, and the start of something new: a UK government backed aggregator?

Generative AI, ChatGPT, and the capital markets

The Cloud Wars have Officially Arrived in the Capital Markets

The SEC Officially Says “No” to Unbundling

A Blockchain Blow-Up You Missed Much Closer to Home (Spoiler: Not FTX)

People are getting fired for using Whatsapp, but no one gets fired for using Email

Wall Street shifts to the Cloud

Asset Manager or Asset Owner – who votes?

Are Research Marketplaces officially dead?

Who will own identity in Capital Markets?