
The hedge fund COO’s job has changed. The operating model hasn’t.
Talk to COOs at multi-manager platforms, multi-strategy funds and single-strategy managers, and the same five pressures come up.
1. Scale without adding headcount. Every new PM pod, strategy, prime broker, instrument or fund vehicle lands on the same operations team. AUM can grow quickly. Ops headcount can’t, and investors and the CFO watch every hire. In SS&C’s 2026 survey of multi-manager funds, 67% said reducing operational complexity was their top infrastructure priority. The trap is false economy: cut the junior roles or delay the automation, and the cost comes back as errors, late nights and risk.
2. Make AI work in production, and defend it. AI is no longer an innovation project the COO can watch from a distance. It’s something they’re accountable for. About three in five of SS&C’s respondents named AI as the trend with the biggest impact on 2026, and nearly half already run it in production. COOs are rightly skeptical of vendor claims. They need to show a board, an allocator’s ODD team and a regulator exactly what the AI did and why.
3. Key-person risk. Every fund has a handful of people who just know how things work. Which prime broker sends the margin file late on a Friday. Which counterparty needs a phone call, not an email. How a particular break was resolved last time. That knowledge lives in their heads and their inboxes. When one of them is on holiday, things slip. When one of them leaves, the firm loses capability, not just capacity.
4. Control you can evidence. Accountability can’t be outsourced. COOs want data lineage, evidence for every action and controls built into the workflow, not checked after the fact. Allocators’ operational due diligence asks harder questions every year, and “we have a spreadsheet for that” is no longer a good answer.
5. High volume days. Most operating models are built for a normal trading day. They fail on the day volatility spikes, a set of counterparties are onboarded, a new instrument is traded, or three people are out. That’s when the margin calls stack up, the breaks age and the risk is highest.
Put together, COOs are being asked to support more business, with more control and more AI, without a proportional rise in cost or risk. The old answer, hiring more people, no longer adds up.
Why the usual answers fall short
Most funds have tried some combination of four things.
- Hire. A good senior ops hire is expensive, takes months to find and longer to get up to speed. It adds capacity but concentrates more knowledge in more heads.
- Outsource. Administrators and BPO providers take volume, but the COO keeps the accountability while losing visibility. Exceptions still come back in-house.
- Build. In-house automation works until the person who wrote it leaves, or the counterparty changes a file format. Solutions are typically narrow and brittle.
- Buy point solutions. Matching engines, reconciliation platforms and dashboards are good at telling you something is wrong. Fixing it is still manual.
Each of these helps at the edges. None changes the shape of the work. A person still has to read the exception, pull the trade from the OMS, compare the numbers, work out what went wrong and write the email.
I’ve seen this play out at clients across geographies and sectors, with critical work happening outside the systems leading to operations becoming a constraint on growth.
A different operating model
For the first time, software can do the judgement-heavy work in between the systems. It can read the email, check the portal, compare the records, work out what went wrong and propose the fix, with a person deciding what goes out.
That changes each of the five pressures:
- Scale. More volume, counterparties and instruments with the same team. Capacity grows without headcount growing at the same rate.
- Responsible AI. A person signs off each action. Every proposal comes with its evidence, and every step is in the audit trail. That’s governance a COO can stand behind.
- Key-person risk. How the firm actually resolves things, today held in a few heads and inboxes, becomes memory the firm keeps.
- Control. Controls are built into the workflow, so the evidence exists before anyone asks for it.
- High volume days. Agents work around the clock and absorb spikes, so the team spends the bad day on judgement, not triage.
The experienced operators don’t go away. They stop being the system and start running it.
What I’d ask, if I were a COO
- Where do our exceptions actually concentrate, and who resolves them today?
- What information would an agent need: which mailboxes, portals and files?
- What level of sign-off do we want on day one, and what would it take to relax it?
- If our two most experienced ops people left tomorrow, what would we lose?
Why now
Three trends are coalescing to make this possible. The foundational model capabilities are good enough to work in production, inside regulated firms provided they are orchestrated appropriately. The cost pressure is real and it isn’t going away. And the talent is scarcer and more mobile than ever.
Every part of capital markets eventually gets its infrastructure layer. The front office got its terminal decades ago, and now every professional depends on it. Operations never did. I think that’s about to change, and it won’t be another dashboard or copilot. It will sit in the middle of the operating stack, resolve the work rather than flag it, and get sharper the longer it runs.
That’s why we started Yantra.
If you’re a COO thinking about what this could look like for your firm, I’d love to compare notes.