Are as-a-service platforms the answer to scaling AI in asset management? Partly. They solve where your systems live. They do not solve who is accountable for what your AI agents decide.
Asset and wealth managers are spending more on technology than ever. Global Tech&Ops spend has reached €173 billion and is projected to hit €243 billion by 2029, according to a recent report by Monitor Deloitte and Objectway. Yet 72% of firms still have not built a truly scalable business model.
For CEOs, CIOs and CTOs, that is the question behind every budget line: why is more spend not buying more scale? And as AI agents for asset management move from pilot to production, a second question follows. When the platform belongs to someone else, who owns the outcome?
Why isn’t more technology spend buying scale in asset management?
The report, published in June 2026 and built on an FT Longitude survey of 300 firms across 10 countries, offers a clear diagnosis: complexity is being managed, not solved. Firms have layered people and technology onto old operating models, and costs now grow almost in lockstep with volumes.
Four findings stand out for asset management leaders:
- Cost-income is stuck. UK wealth managers and private banks reached a 78% cost-income ratio in 2024.
- Regulation is now a technology bill. Around 65% of the EMEA compliance budget goes on technology, and roughly 30% of firms cite regulation as the main barrier to AI ROI.
- The front office is hardest to scale. Client-facing workflows depend on coordination and human judgement, which is exactly where automation has stalled.
- Agentic AI is accelerating. Deployments are expected to grow about 2.3x from 2025 to 2026, and AI use in portfolio construction is set to rise from 5% to 25% of firms.
The report’s answer is as-a-service sourcing: SaaS and BPaaS models that turn fixed costs into variable ones. Their share of Tech&Ops spend has already risen from 8% to 18%. In one case, a mid-sized Italian asset manager cut its Transfer Agency and NAV team from around 13 people to about 4, who now focus purely on oversight.
We agree with the diagnosis. We take a different view of the cure.
Our view: you can outsource the platform, not the accountability
We believe the most important idea in the report sits almost at the end: firms must distinguish systems of record from systems of action. That distinction, not the sourcing model, is where the next competitive gap will open.
- Systems of record are portfolio accounting, Transfer Agency, NAV, custody and tax reporting. The rules are the same for everyone. Here as-a-service makes sense, and we would encourage most firms to buy rather than build.
- Systems of action are the AI agents that read research, draft proposals, check suitability, monitor compliance and prepare client reporting. This is where judgement lives and where differentiation is built.
From our experience three things follow.
1. Value compounds across workflows, not inside one vendor. Speed up one step and you get a gain. Speed up research, proposal, compliance check and reporting together and you compress the whole cycle. Those steps typically span four or five different platforms. No single provider will orchestrate them for you.
2. Accountability does not transfer with the contract. Under DORA, a firm remains fully responsible for its obligations when it uses ICT third-party providers. The EU AI Act places duties on deployers, not only on providers. Supervisors in Dublin, Luxembourg and Zurich expect the firm itself to explain what an agent did and why.
3. More vendors means more need for an independent control layer. As AI arrives embedded in every SaaS platform, each with its own logic, someone has to watch all of them: audit trails, explainability, drift, incidents. That layer cannot belong to any single vendor. It has to belong to you.
Put simply: you can rent the factory. You cannot rent the licence to operate.
What should asset management leaders do now?
1. Map your systems of record and systems of action
Before any sourcing decision, list your core workflows and mark each step. Standardised, rule-bound steps are candidates to buy as a service. Judgement-heavy steps that touch clients or portfolios are where your own agents create value.
2. Redesign the workflow before you migrate it
The report warns that firms which ignore AI during their SaaS transition will execute the same transformation twice. We agree. Measure the redesign with operational metrics: cycle time from research to client proposal, share of steps automated with human oversight, and incident rate.
3. Build an AI governance layer above your vendors
Keep a single inventory of every AI agent you run, including those embedded in vendor platforms. For each one, define the audit trail, the explainability standard and the point where a human signs off. AI agent compliance monitoring should work the same way whether the agent is yours or your provider’s.
4. Turn your people into supervisors, not spectators
The Italian case is instructive: fewer people, but in a stronger oversight role. Agents should prepare the work and humans should own the decision. That is how you scale without losing the trust of clients and regulators.
Conclusion: orchestrate, don’t just outsource
As-a-service adoption will keep growing, and for systems of record it should. But the next gap in AI in asset management will not be between firms that bought SaaS and firms that did not. It will be between firms that can govern their agents across an entire ecosystem and firms that cannot.
AI governance for asset managers in Europe is not a brake on that journey. It is what makes the journey possible.
The winners won’t be the firms with the most platforms. They’ll be the firms that know exactly what their agents are doing, and can prove it.
Source: Monitor Deloitte and Objectway, “Master complexity to unlock scale: the as-a-service imperative for Wealth, Banking and Asset Management”, June 2026.


