Navigating the middle office maze: Mizuho and Capco on transformation, tech and control
We brought together two middle office experts for a Lunch & Learn session at SIFMA Ops 2025. Mizuho managing director Claire Harrison and Capco managing principal Stephane Ritz joined our chief product officer James Maxfield for an insight-packed discussion.
Their conversation covered everything from the challenges and opportunities for middle office transformation to the potential and pitfalls of AI. They also discussed how to maintain control in today’s fast paced market environment and keep hold of the best Operations talent.
Let’s dive into some of their unmissable insights.
1. Challenges and opportunities for middle office transformation
“Everybody talks about back office transformation, they talk about reg transformation, but no one really talks about the middle office,” James began the discussion.
There are a couple of reasons why it’s difficult to drive transformation in the middle office.
“You’ve typically got multiple stakeholders,” he explained, “so different business units, different regions, different entities. That makes it quite hard to get consensus around decision making or decisions around investment.”
On top of this, systems are normally fragmented. It can be hard to drive standardisation and consolidate processes when the equity business uses a different system to the fixed income business, who use a different system to the treasury business.
But there are many opportunities to transform the middle office for greater efficiency and agility.
The automation, enhancement and improvement of point-to-point reconciliation controls really can’t be underestimated, Claire said. She notes the potential for machine learning to interrogate documents such as confirmations or prospectuses to automate the extraction of key information.
The key is “to really help people do more with less. Because we know in Operations every time volumes go up, we don’t go up in staff, so we have to continue to manage with the same number of people.”
Stephane also agreed that the opportunity lies in automating tasks that people have to do anyway, giving the example of month-end, quarter-end or year-end closes. “The industry standard is about five days to close the books and a lot of firms have been trying to close those much faster.”
“The beauty of automation is that it should facilitate a lot of it. But even outside of reconciliations you do have the substantiation process as well. That’s a much more time consuming process, but automation or GenAI could actually help with the narrative or the commentary on those breaks.”
“That’s always been something very interesting and challenging for financial organisations – the connection between middle office and back office; your subledger to general ledger reconciliation. For some reason that never goes very smooth. It’s very labour intensive.”
2. Current market conditions are increasing the pressure on middle office teams and processes
A lack of automation causes problems for the middle office all year round, but current conditions are putting extra strain on processes, teams and operating models.
The market has seen extraordinary volatility so far in 2025. The VIX – also known as the ‘fear index’ – breached 60 in early April for the first time since the early days of the Covid pandemic. A few days later, the S&P 500 posted its biggest one-day gain since 2008, with a record 30 million shares being traded.
So what impact do conditions like this have on the middle office? “The biggest concern when there’s massive periods of volatility is your primary stakeholders in the middle office are your front office,” Claire said. “They want everything complete, accurate, perfect all of the time. And as soon as they’re putting a huge amount of volume through the system, you’re focusing on that. Your capacity is taken away from your control environment because they just need to be focused on the trades.”
The heightened volatility is placing greater pressures on operating models. Many of these are fragmented across the globe, whether due to cost, scale or capacity issues. One of the challenges this creates is around standardisation. How can firms maintain consistency?
Stephane said that firms have to strike the right balance between standardisation and flexibility. It’s not possible to standardise everything – for instance, regulatory compliance differs from region to region.
“There is a little bit of harmonisation between regions, even though you’re going to see different regulatory requirements,” he said. “You can take the EU versus the US: collateral management, margin, etc. Those regulations are somewhat different, requiring different calculation engines, different reporting. The goal is to maximise that standardisation, but there will be some local Centre of Excellence still in place to make sure that you remain compliant.”
Financial firms have long pursued the goal of standardisation. But, in fact, this can create its own problems. “In the effort to standardise, you’ve almost encouraged bespoke and nonstandard products,” James said. “Because obviously that creates more opportunities and more money.”
“You can make more P&L off one fun bespoke trade than you can a couple of weeks’ worth of very standardised trading,” Claire agreed.
And these bespoke products require their own workflows to manage, causing a proliferation in non-standard processes.
3. Artificial intelligence needs a careful and considered approach
AI was, of course, a topic that came up frequently during the conversation. The panel agreed that AI is often sold as the cure to all problems and that firms need to take a more considered approach to get the most out of the technology.
“Everybody’s got different issues and challenges, so it’s really a customised approach to the client,” Stephane said. “But we are trying to promote the quick wins. Your pre-trade type of activities, your post trade type of activities, your KYC (Know Your Customer). Things that actually have an impact on your STP (straight-through processing) footprint in the organisation, those are your quick wins and your best candidates. But depending on where you are in your journey of automations and legacy refresh of your platforms and hardware, the recommendation is going to be slightly different.”
This approach allows you to prove the concept and show value, without spending too much money, energy and resources on a project that doesn’t produce the right benefits.
It seems like an obvious point, but the key to winning with AI is to figure out where it is going to deliver the most benefit. “There is that risk of trying to use those new gadgets and over-automate or overthink,” Stephane explained. “Next thing you know you’ve got hundreds of POCs (Proof of Concepts), but nothing is productionalised or scaled.”
“Drive your pilot and make sure that it’s in a place where you can productionalise and release your solutions.”
On top of focussing on delivering value, firms should make sure that everyone understands what’s going on and what the tool is doing.
“To me it makes sense to start from the beginning and focus on the bigger volumes, the bigger areas,” Claire said. “Get it really working so that people understand it, we know how to control it, we know how to tell the regulators we know how to control it, then start to move to the more complex.”
“Right now everybody’s excited because they’re playing with the AIs, the co-pilots, the things that used to take you an hour to write your notes up during a meeting. It’s printed in your email in two seconds. There’s some real gain in that, but there’s also some risk with that.”
In particular, there’s a risk of losing operational knowledge when people don’t get the chance to learn, because AI and automation is doing everything for them.
“It’s trying to balance that experience, but also make sure that people still understand the core of our business with all of this automation and all of this STP,” Claire explained. Those bespoke products mentioned earlier can require equally bespoke processes.
“You have to use your skill set, your knowledge, your experience to be able to work out how to put that in an environment and make sure it’s properly looked after. Where people are coming up in these very STP automated environments, they don’t have that problem solving mindset and trying to incorporate that within our environments is one of our key challenges.”
And that organisational knowledge is vital in enabling AI to do what it does best, Stephane added. “We are seeing a lot of GenAI in terms of building new platforms, whether it’s a clearing platform, settlement platforms, etc. But again, that comes with a lot of knowledge, making sure that you’ve got the right knowledge base so that your platform that you’re building is actually the best possible one.”
He also noted the need to ensure the necessary level of control around your automation. “We still suggest our clients continue to have the right level of controls. Automation and GenAI doesn’t replace a fully controlled environment. There’s always some type of human factor to it, from a regulatory perspective.”
This is where the middle office really needs to own its skillset and role within the organisation, Claire said. “The key thing there is to remember who and what middle office is supposed to be. We’re professionals in the Operations division. We are control specialists. And the change governance in that control environment cannot be day two. It has to be there on day one and it has to be right on day one.”
“It’s making sure that people understand the risk and the cost,” Stephane added.
4. Retain top middle office talent by improving the employee experience
The previous section touched on the fact that new Operations talent risk missing out on vital knowledge. This is something firms need to address because specialists are a rare commodity in the industry.
The middle office skillset is very difficult to find in the market, James said. This is complicated further because processes and systems vary from firm to firm – a subject matter expert coming from another firm may not understand your way of working.
“To me, retention is a couple of things,” Claire explained. “One of the key things is that people can see a future. They can see that they’re not stuck in a rut, they’ve got an opportunity or if they like staying at a certain level of contribution that it can work.”
The latter part is important, she said, because not everyone aspires to climb the career ladder and – quite frankly – organisations don’t have the capacity to elevate every single employee.
“But even people who don’t necessarily want to run the division, they may want opportunity to learn a little bit more and step a little sideways. One of the things that we’ve seen with really good success is to have Centres of Excellence: a combined team that have a little bit wider footprint and it gives people opportunity to learn without necessarily having to step up.”
This gives people the opportunity to broaden their skillset – moving from OTCs to futures or securities, for example – while at the same time having a closely integrated organisation that works together. “That helps us keep people,” Claire said. “And also it means that that information is now kept within the organisation. That SME [subject matter expert] knowledge isn’t walking out the door.”
Stephane agreed that Centres of Excellence are “probably one of the best tools” to retain talent.
5. How to get shareholder buy-in for transformation
So, the need for transformation is clearly there. Firms are struggling with outdated systems, fragmented operating models and an unscalable network of people. And the rewards of improved efficiency and control are significant. The firm gains a competitive advantage from improved client service and the risk of regulatory penalties is reduced. They create a better experience for their employees, enabling them to keep hold of valuable talent and expertise.
But how do you convince your stakeholders of all this?
“The key thing I’ve learned is you have to appeal to their own interests,” Claire said. “Most of the time we’re bidding against the front office for some spend and we have to persuade them the spend that we want to make in the middle office and back office space is valuable to them.”
One strategy is to show them how much a lack of automation has cost them in the past. For instance, a ‘fat finger’ error or an error that wasn’t noticed for three weeks.
“This is important because that completeness and accuracy, if it’s wrong, we all suffer.”
On top of cost, there’s also the risk and compliance elements, Stephane added. “You don’t want to put your organisation at risk with regulatory findings, fines, etc.”
“The risk and control angle is important because obviously organisations can lose money,” James added. It’s more than just a trader taking a hit to P&L, however. “Increasingly we’re seeing regulators now sanctioning people for poor control environments.”
The risk of regulatory fines definitely helps the business case, Claire said.
“This is an absolute requirement. I have to be able to take your trades from the front to the back correctly, properly, completely and accurately. Because otherwise it impacts the balance sheet, impacts the P&L, it impacts the customers.”
The client service aspect is also an important selling point, she added, especially in today’s highly competitive environment where price isn’t much of a differentiator.
“If we’ve got a better control environment, a cleaner control environment, we catch the errors on day one, not on day 15, the clients are happier. The clients do come back to you if they get a better service. So if there’s a tiny, tiny margin between us and another thing, but the Operations function is way better, we’ve heard from clients that they will choose to prioritise trading with us over someone else.”
Ultimately, it’s about packing all these different factors together so that stakeholders with control of the budget want to spend it on automation and streamlining.
Conclusion
At the end of the discussion, James asked our panellists what their one takeaway would be for the audience.
“Make sure it works for you,” Claire said. “Don’t be persuaded to take something on that doesn’t fit your need just because it’s pretty.”
“Middle office automation is not just about cost reduction,” Stephane said. “It’s really about creating a resilient, compliant and optimised operating model. That’s really what it is.”