I suspect this is one of the harder decisions facing asset management COOs over the next few years. Identifying something that could theoretically be automated is relatively straightforward. Understanding whether changing it actually saves money, improves the business and leaves the firm with an acceptable level of operational risk requires considerably more thought.

In-house development is becoming more interesting as part of that discussion. AI is allowing capable people to create useful tools more quickly and at a lower initial cost, particularly where a firm has a specific problem that isn't well served by an off-the-shelf product. There will undoubtedly be cases where this makes good commercial sense, but the full cost needs to be understood.

An external licence may look expensive, but somebody else is responsible for maintaining the product, developing it, documenting it and making sure it continues to work. Bring that capability inside and some of those responsibilities come with it. If the person who built an important internal tool subsequently leaves, the firm needs to know that somebody else can understand and maintain it. Asset managers have thought about key person risk within investment teams for years and may increasingly need to apply some of the same thinking to technology and operations.

There is a similar issue with calculating the return. Removing a software licence produces a saving which is easy to see. Additional internal development time, oversight, maintenance and operational risk are harder to quantify, particularly if they sit across several budgets. A cheaper solution is only cheaper if the whole cost of running it has been considered.

All of this makes the quality of the operations leadership particularly important. A good asset management COO has always needed to understand people, processes, providers, risk and the requirements of the investment business. Technology and data are becoming a larger part of that responsibility. I don't think a COO needs to be a developer, but they increasingly need enough knowledge to challenge the people advising them and to understand the commercial and operational consequences of the decisions being proposed.

The skills required underneath the COO are changing as well. If technology gradually removes more repetitive work, there should be greater value in people who understand why a process exists, can investigate problems and are comfortable working across operations, technology, data and investment teams.

Deep knowledge of a particular system will remain useful, although firms may also want people who can question whether that system or process remains the right answer.

This creates an interesting issue when recruiting a COO, Head of Operations or senior operations professional. Historically, firms have often placed considerable weight on experience of their particular operating environment and technology. That is understandable and sometimes essential, particularly where somebody needs to become effective quickly. The difficulty comes when a firm wants to change its operating model but recruits somebody principally because they have spent their career working within something very similar.

The better starting point is probably to understand what the business expects to change over the next few years. A firm undertaking a substantial transformation may need somebody who has redesigned an operating model before and knows what tends to go wrong. A business created through acquisition may place more value on integration experience. A smaller asset manager may need somebody comfortable with a broad remit, while a larger business can afford considerably more specialist leadership.

There is also the existing team to consider. Asset managers have spent years building operations functions around the technology and processes they needed at the time. As those requirements change, some people will develop with them and others may have experience which becomes less central to the future model. That is a normal consequence of businesses changing, and hiring externally should not automatically be the first answer. Someone who understands the firm deeply and can acquire the additional skills may be far more valuable than an outsider who arrives with a fashionable technology background but little understanding of the business.

Contractors can be useful here as well. Transformation and integration often create a temporary concentration of work and firms do not always know at the beginning what the permanent structure will look like once the project is complete. Bringing in somebody who has dealt with a similar problem before can add experience and capacity without forcing the firm to decide the long-term organisation too early.

Where an external appointment is needed, the search increasingly needs to get underneath job titles and lists of systems. Knowing Aladdin, Charles River or another platform may be important, but it tells you relatively little about whether somebody has improved an operating model, managed difficult change, dealt with resistance internally or knows when leaving something alone is the better decision.

That last quality may become particularly valuable. There will be plenty of enthusiasm around AI and plenty of technology available to asset managers promising lower costs and greater productivity. Some of it will undoubtedly deliver both. Some of it will simply add another system, another licence and another layer of complexity to an operating model that was already complicated.

For CEOs and boards, the quality of the people making those choices will matter. Operations leaders will increasingly be expected to understand the existing infrastructure, make sensible use of AI and data, manage risk and work out where genuine efficiencies can be achieved without disrupting the parts of the business that already work well.

That is a broader brief than simply running operations, and I suspect it will gradually change what asset managers look for when they recruit COOs, Heads of Operations and the teams around them.