More Intelligence Needed in Asset Management

Everyone knows the robots are taking over the world. British Airways’ High Life says so, the Daily Mail says so, and to prove asset management is keeping up, I’m convinced I saw a lorry delivering C-3PO to the FCA’s London office just last week. 

I clearly jest¹. But the deceptively simple questions I’m often asked include: 

• Where can artificial intelligence (AI) actually help my business generate investment returns? 
• It’s all science fiction, isn’t it? 
• I’ve spent all my budget on regulatory issues/fixing my legacy tech/McKinsey, and isn’t it really expensive to properly test AI? 

As a decades-long student of industry trends, what I care most about is helping clients improve their business and giving them sensible answers to sensible questions. Given that, I think it’s critically important to put AI in context and add some classic intelligence before I come back to the artificial bit. I’m not going to waste time describing the ins and outs of AI, not least because there are far better folk to do that than me (Data Science Central or Introducing LLaMA to name two]. With that as context, let’s explore. 

The big challenges demanding our focus 

I liked Northern Trust’s recent report “The Evolving Asset Management Landscape: Only the Fittest Will Thrive” which presented findings on how organizational leaders are adapting their operating models for growth. The survey respondents included 151 asset managers across the globe with AUM ranging from $250M-$150B+. 

Over the next three years, the respondents rated their top internal challenges as performance (59%), talent management (50%) and rising costs (44%). Another reason I liked this survey was that it appears to be much less swayed by operations and tech folk bemoaning their lack of success in data management. In contrast, it was centred around the business needs; the main job of an asset manager is to provide good performance, after all.  

Given the survey’s identified challenges noted above, the top four strategic responses are to: 

1. Launch new products or investment strategies (52%)
2. Cut costs and increase efficiency (43%)
3. Improve sales and distribution (37%) 
4. Focus on increasing market share in current products (37%) 

In thinking about these four strategic priorities, I realized that they can be grouped into two neat streams: 

STREAM A: Clean up BAU and just get the old products off the shelf now, please. (Priorities 2 + 4)  

STREAM B: Deliver better performing, more suitable and accessible investment products to clients (Priorities 1 + 3) 

So recalling my original questions, the practical answers to “where can AI actually help my business?” should address the strategic priorities above. Let’s look at each in turn and see if robots can (or cannot) help. 

Stream A: Clean up your BAU 

A whopping 63% of respondents indicated that in planning for efficiencies and cost savings, they would be deploying new technologies. In contrast, implementing a new target operating model (TOM) was a lowly 22%. At first glance, that sounds counter-productive: your current tech is underperforming so just *add* *some* *more*! But, I reason that a new TOM is hard and time-consuming and bolting in a new tech is much, much easier. Adding AI capabilities to improve efficiency seems like a credible approach. Some recent examples of AI creeping in and enhancing the current capabilities include: 

• BloombergGPT is a Large Language Model (LLM) trained specifically on Bloomberg data meaning queries, news searching and sentiment analysis should be faster and better than ever 
• Several AMs (such as JPMorgan, Schroders & PIMCO) have been building out their own ChatGPT clones to help with fund commentary, research, and analysis 
• SS&C’s Blue Prism combines robotic process automation (RPA), business process management (BPM) and low/no-code to deliver their updated version of intelligent automation (IA) services 
• The leading cloud providers (AWS, Azure & GCP) are all promoting their AI offerings: enhancing analytics, increasing efficiency and allowing firms to build and deploy their own AI solutions.  

Stream B: Deliver Better 

Back to stream B and delivering better performing products to clients. I think this is the challenge that has been plaguing the industry forever. And I think it’s because our industry would rather focus on products over clients, data over needs, the numbers over people. The vast majority of tech development to date has driven improvements in portfolio management, electronic trading, compliance, IBOR, and post-trade outsourcing, none of which touch the client directly.  

The biggest and most opportunistic irony for asset managers is that we can now leverage the robots to be more human(e) and more in-tune with clients than we’ve ever been. 

Many other sectors have been doing this for decades. The humble grocer (in the form of Tesco or Walmart) knows more about their clients than any of the highly sophisticated asset managers I know. With AI, we can start to really understand the shifting needs of our clients transparently, ethically, and probably in real-time. And yes, this transcends the internal definitions of retail, institutional, wholesale, wealth, etc.; they are all clients and AI can help us understand them and service them better than ever. Now, that’s truly exciting. 

So, in reference to the second simple question at the beginning of my blog: No, it’s not all science fiction, not at all. Just take a peek at toggle.ai which is a startup that helps everyone invest smarter by teaching ChatGPT to invest. An easy idea that lowers the barriers to client engagement. AI is happening now and I guarantee that the big players in the market have been trialing, testing and delivering AI-driven initiatives already. Small, yes, imperfectly formed, yes, but we all know what happens to ugly ducklings that get looked after properly.  

And now we come back to the last simple question our clients ask: how do I get into the AI game with no budget? 

Budget isn’t the issue, nor risk so get busy 

As another context setter, Tesco’s revenue for FY22/23 was £65,762M and its statutory profit after tax was £753M giving a crude profitability of 1.14%. In contrast, a small/medium-sized UK asset manager’s recently released results state a net revenue value over 350 times smaller but a similarly calculated profitability of 19.2%. There are many managers with similar or better performance. Even as a much smaller business, this asset manager’s profitability is tracking at over 16 times that of one of the UK’s leading retailers. Asset managers have budget. That’s not the problem. 

Imagine, just imagine, what asset managers could do with their clients if they decided to invest in them just like Tesco. They have over 16 times the relative profit % to start with. Asset managers have all the funds necessary to be market leaders in client knowledge, client excellence, and client value. 

Secondly, it is easy to ‘properly test AI’. With countless fintechs promoting their AI offerings, the problem is not in the supply either—it’s with prioritisation. The trick is to be clear in your business strategies, in understanding what a good outcome looks like, and over what timeframe. There’s no substitute for experience, and I’ve been involved in countless proof of concepts, prototypes, MVPs, MVSs and DeLoreans to know what good looks like. Of course, you need good ringfenced data, business stakeholder buy-in, and the time with passionate, open-minded people to realize change.  

The last thing I would strongly advocate is to test out AI with as many proactive and trusted third parties as you can manage. External collaboration helps you to sense-check your ideas, ensures you are crafting viable solutions for your market, and leverages collective insights from solutions providers and peers. Too many wildly impractical ideas work in a vacuum; we have to allow new ideas to struggle in adversity and be assailed from all angles. Our industry needs resilient solutions, not fragile facades. 

Using robots to focus on people: what would C-3PO do? 

Artificial intelligence is nothing without our own intelligence. If we ask silly questions, it’ll just amplify that in its answer—remember how a great amp distorted a poor cassette? (Showing my age—sorry experience—again). But if C-3PO really were powered up, he’d politely recommend this:  

Erase poor performance.

• Stream A: remove the bad funds, rationalise funds with no excuses, strip operating model to the leanest and meanest 
• Stream B: invest in the future, in new asset classes, in digital assets and quickly learn where new alpha can be uncovered in a crowded market 

Erase poor client service.

• Stream A: drain up review with client feedback, steal the best ideas from outside the industry, compensate for prior bad service 
• Stream B: collaborate with the market, proactively collaborate with your clients, radically refocus on what all clients want 

Erase unsuitable products. 

• Stream A: reduce fees, re-evaluate pricing approaches, optimise distribution 
• Stream B: take Consumer Duty’s principles to heart, help design ethical asset management AI, work with and learn from fintech by default 

We’re here to utilise all types of intelligence in that pursuit.  Yann LeCun, Chief AI Scientist at Meta, said that “our intelligence is what makes us human, and AI is an extension of that quality.” Pooling and aligning all that intelligence seems to be the only smart thing to do. Even C-3PO once warned “R2D2, you know better than to trust a strange computer!”