This page is an example of how we run an AI advisory retainer for a wealth manager, and it does not describe a client.
The situation
A wealth manager gets a steady stream of AI vendor pitches: meeting note takers, tools that draft statements of advice, portfolio commentary and client chat. Nobody on the leadership team can judge them on the technical detail. That detail is where client data goes, which model sits underneath and what the vendor logs and keeps. Some staff already use AI tools, and the board asks how the firm manages that risk. The firm needs a standing technical opinion rather than a project.
What we build
We build no system in this engagement. It runs as an advisory retainer, with Zinto on call to the leadership team and the board.
Vendor and tool reviews
Before the firm signs, we read the documentation and the contract, put technical questions to the vendor and write a short assessment. It covers where data is stored and processed and whether client data trains the vendor’s model. It also covers the access controls, the logs and how the firm gets its data out at the end.
Build versus buy
A vendor tool is sometimes close to what the firm needs but not a full fit. We then compare it with a build in the firm’s own cloud account. An example is retrieval over research notes and client file notes with an open-weight model such as Llama or Mistral. The comparison covers the cost to run, the cost to change and who holds the data.
Governance and model risk
We help the firm keep a register of the AI tools in use. Each entry names an owner, the data the tool touches and how the tool is tested. We give input on the model risk policy and review a tool again when its vendor changes the terms or the model underneath.
Board papers
We write or review board papers on AI, which cover what the firm uses, what it plans, the risks and the controls. Directors can put questions to us before or at the meeting.
What your team owns
Leadership and the board own every decision, and our input is written advice. The register, the policies and the board papers belong to the firm and stay in its systems. An assessment can show that a build is worth doing. That build becomes a separate project with its own scope and price.
What to expect
The board gets a technical assessment before the firm signs with an AI vendor. The leadership team gets a written comparison of build and buy when the choice is close. The firm keeps a current register of the AI tools it uses and who owns each one. Questions between reviews come to us, and the answers draw on what we already know of the firm’s systems.