Transparency is the key to evaluating AI solutions for the buy-side

INDATA’s David Csiki discusses the advancement of AI on the buy-side and how investment management firms can effectively evaluate AI-based solutions.

AI is everywhere. From a media standpoint, one day AI promises to transform the way people work and live, from doing their jobs faster and more accurately, to curing diseases like cancer. The next day AI promises to put people out of work or, even worse, put an end to humanity itself.  Such extremes make it difficult for reasonable people to form measured and rational opinions of AI with the risk of many folks just “tuning out” the subject of AI altogether.

When it comes to the buy-side, tuning out AI is no longer an option with more and more vendors announcing their own AI-based solutions promising to revolutionise research, trading, compliance, portfolio management and operations.  In fact, research firm Gartner predicts that, “as AI technologies advance, they will reshape how investment management firms operate, communicate and make investment decisions.” Source: Gartner: Top 10 Technology Trends for Investment Management CIOs in 2023.

Now more than ever, investment managers need to think about how AI can be of benefit to their own firms with the goal of future-proofing their key software systems and processes.

Start with a general understanding of AI

As much as AI is continuously being discussed in the press, what does it really mean? As a SaaS provider of front-to-back-office solutions including trade order management (OMS) and portfolio management, accounting and reporting, at INDATA we spend a significant amount of time speaking with clients on AI and how it can help solve their own unique business challenges via our solutions. To this end we have published a resource Future Proofing the Front Office: AI | INDATA and also a whitepaper Future Proofing the Front Office which provides a background on AI as well as the other technology trends shaping the buy-side. 

It is important to discern the subfields of AI such as generative AI, NLP (natural language processing), machine learning and deep learning. Understanding these aspects of AI is essential to form an intellectual framework on how AI might be of benefit to one’s own firm.

Think about inefficiencies and how AI might be able to assist

Once one has a general understanding of the different subfields of AI, the process of beginning to think about aspects of the business of investment management that might benefit from various AI approaches can begin. For example, in the area of investment research generative AI solutions such as Chat-GPT have the potential for allowing analysts to consume and synthesise large volumes of information much more quickly, which would not have been possible previously. While on the trading desk, machine learning via trading algorithms has already been in place for a number of years via broker sponsored trading algos and front ends. 

Other departments within investment firms that now stand to benefit from AI are in the areas of compliance, client servicing, marketing as well as making the portfolio management and trading workflows used to create and execute trades more efficient in themselves. This last area is one that we believe holds the most promise because the workflows contained in many systems have become inefficient and legacy-based and AI can do much to make more of a straight line out of convoluted processes.

Focus on “practical” AI and avoid vendor hype

In addition to speaking with clients, we are asked from time to time to comment on other vendor product offerings involving AI. While each vendor has their own take on AI within their solution, we have found that what works best is to offer practical solutions with tangible outcomes rather than general or overly ambitious goals that are possible to achieve with AI. At present, several vendors are jumping on the generative AI bandwagon, however, one needs to understand the associated risks with generative AI in that results can be spurious or inaccurate altogether.  Such phenomena, often called hallucinations, contain inherent risk, and should be carefully considered when evaluating generative AI solutions.

Another thing to consider is whether vendor-based AI solutions are in production and, if so, for how long. Often vendors will launch an AI beta hoping to gain traction and build use cases with clients. Telltale signs of this include broad-sweeping marketing language used surrounding AI generated “insights” without offering specific examples and vendors that push products out for “future” release to “select” clients.

Only evaluate AI solutions that offer full transparency surrounding technology, methods, and possible outcomes

Once an area of interest has been assessed and approved as being fit-for-purpose and available vendor solutions have been reviewed, the next step is a deeper dive into the vendor’s proposed AI solution. Questions to ask include what is the vendor’s technology stack with regards to a given AI solution? How is the AI component integrated into their solution? What safeguards has the vendor put in place surrounding the output generated by the AI component?

Asking the right questions will make any subsequent trial or pilot of an AI solution that much easier. For example, if the investment manager learns that the vendor’s technology stack is not compatible with their own technology stack (i.e., Microsoft vs. Google), then no further action may be warranted given the difficulty of integrating differing platforms for a set of outcomes that provide for meaningful ROI. Additionally, if the vendor cannot fully explain the technical aspects of the AI solution or if the AI component has a “black box” approach then no further action should be taken. Lastly, if vendors do not have safeguards put in place surrounding their AI component, then this is a red flag that a specific solution may not be a good fit given an investment firm’s compliance requirements and fiduciary responsibility to clients. In summary, vendor transparency is the key to evaluating AI solutions on the buy-side and should be the starting point for any firm looking to implement an AI solution for their own unique needs.

To learn more about how INDATA utilises practical AI within its SaaS-based solutions for OMS & portfolio management, reporting and portfolio accounting please visit