Agentic AI in Private Banking: Why Source of Wealth Is the Real Test
CCO Managing Partner
The first wave of AI in banks lived in the inbox. A better email. A drafted note. A summary of a meeting nobody wanted to write up. That wave was easy to like and easy to ignore. Nothing fundamental in the control framework had to move. Nobody had to change who signs.
The second wave is different. It is moving into onboarding, Source of Wealth, KYC, AML and periodic reviews. These processes combine high volumes, heavy documentation and, at the end, human judgment. That is exactly why they are strong candidates for agentic AI. It is also why they will go wrong if banks optimise only for speed.
Deutsche Bank Private Bank’s new Source of Wealth agent in Singapore and Hong Kong is a good example of where this is heading. Research, documentation and preparation can move to the agent. Human review and accountability stay with the bank. That distinction matters. Tasks can be automated. Accountability stays with people. I would print that sentence above every agent project in a regulated institution.
Why Source of Wealth Is Different
In my banking years, a Source of Wealth file was rarely one document. It was a company sale in one country, dividends in another, a trust the client mentioned late and three versions of the same story sitting in three systems. The relationship manager spent the week collecting evidence. Compliance spent the next week asking for some of the same evidence again. The client only felt the delay.
That process is still one of the most honest tests in wealth management. Source of Wealth is not simply a form. It is the moment the bank has to decide whether it actually understands how the money was made and whether it can stand behind that understanding years later.
Most onboarding work is administrative. Source of Wealth is not. It sits at the intersection of three things banks rarely hold in one place:
- Facts that can be checked: share registers, sale contracts, tax statements, public filings.
- A narrative that has to make sense: where the wealth came from, how it moved and what is still missing.
- A judgment that cannot be outsourced: is this origin plausible, complete and acceptable for this institution?
The easy files look simple. A long employment history. A clean inheritance. A documented exit from a listed company. Those cases should never have consumed a week. The hard files are the ones that define private banking. Owners with several entities. Wealth accumulated across jurisdictions. Structures that were legal, messy and poorly documented when they were created. That is where the biography either holds or the relationship starts with friction.
What Agentic AI Can Usefully Take Over
Source of Wealth is a strong use case for agentic systems precisely because so much of the work is assembly, not advice. A capable agent can:
- Pull approved public records and match them to documents already in the file.
- Draft a first narrative of the wealth origin instead of starting from a blank page.
- Flag contradictions: dates that do not line up, entities that appear only once, amounts that cannot be reconciled.
- Keep one version of the story across onboarding, compliance and the next periodic review.
- Reduce the loop in which the same question is asked three times by three departments.
That is not a small gain. It is the difference between a banker spending the week as a document hunter and spending it as the person who still has to look the client in the eye.
It also addresses one of the worst parts of onboarding from the client’s perspective. Clients do not celebrate a better internal workflow. They notice how often they are asked for the same document, how long it takes before the relationship can start and whether the banker in front of them already knows their story or is discovering it during the meeting. Onboarding is the first proof of whether the institution is organised around the client or around its own departments. Source of Wealth is perhaps the sharpest version of that proof.
What Cannot Move With the Prompt
Speed only helps if the same story survives the first serious review. An agent can collect evidence. It cannot decide whether the origin of the wealth is understood well enough for this bank. It cannot weigh the conversation in which the client finally mentions the trust. It cannot carry the accountability when a regulator or internal audit asks how the conclusion was reached.
That split is the operating model. Tasks can be automated. The conclusion still needs an owner. The client still needs one person who can explain the file without reading it aloud from a chat window. If that ownership is only formal, you have not improved Source of Wealth. You have industrialised a weak biography and given it more confidence.
The same applies to periodic reviews. If the annual update produces a new version of the client story every year, you never had a knowledge layer. You had a writing contest. The narrative that leaves onboarding should be the narrative that arrives at the next review, updated where facts have changed but built on the same owned foundation.
The Hard File Is the Real Test
I would not start a Source of Wealth pilot with salaried profiles. Those will look good first. I would take the hard file. Several structures. Several jurisdictions. Wealth that did not arrive as a payslip. A story that has accumulated over decades rather than arriving neatly packaged in one document.
Then I would ask four questions:
- Did the agent produce one consistent narrative, or three fluent ones?
- Did it surface the contradiction a human might have missed, or bury it in polished language?
- Can the banker explain the conclusion without the prompt history?
- Would the client recognise their own story in the file?
If the answers are yes, this is no longer an interesting experiment. It is a better way of working. If they are not, you have a nicer pilot and the same old biography, only written faster.
The Operating Model Hidden Inside the Workflow
Regulated work has always had three layers. Banks could afford to mix them up because experienced people were often able to bridge the gaps: facts that can be checked, a narrative that has to make sense and a judgment that cannot travel with the prompt.
Agentic systems can be very useful in the first two when the sources are clear and the definitions are shared. They become dangerous in the third if the human signature is only a formality.
That is why this is not really a chatbot use case. It is a workflow and operating-model use case. The question is not whether the agent can produce a good Source of Wealth narrative. The question is whether the bank can define which parts the agent may assemble, which exceptions it must surface and which conclusion still requires a person who understands the client and can defend the decision.
If three departments still keep three versions of the client, the agent will not fix that. It will scale the disagreement and give it a confident tone.
What I Would Tell a Swiss Private Bank
Do not start with a copilot in Outlook and call it transformation. Start with one regulated workflow where the pain is real and the owner can be named.
Put the agent on research, matching, drafting and exception flags. Keep the conclusion with a person who can be asked how they reached it. Make the narrative inherit. The story that leaves onboarding should be the story that arrives in the next review. The same facts, ownership and reasoning should travel with the client instead of being rebuilt by the next department.
And measure the outcome from the client’s perspective as well: fewer repeated requests, shorter onboarding, fewer internal loops, a relationship manager who knows the story before sitting down with the client.
The inbox made AI look friendly. The regulated workflow will decide whether it is actually usable. Tasks can move. The file still needs an owner. Speed is only progress when the same story survives the first serious review.
If you are exploring where agentic AI could take real work out of KYC, Source of Wealth or another regulated banking process, book a 30-minute exploration call with us. We can look at one concrete workflow, separate the work that can safely move to an agent from the judgment that should remain human, and identify where better orchestration could improve both efficiency and the client experience.