Agents are only as good as what they know
Marketing leaders in financial services are already running AI agents somewhere with impressive tests but very few are seeing the payoff at scale.
They sign off a promising pilot, an impressive demo, then it ends up in the compliance queue. The technology works – the business case stalls.
Agents should be thought of as brilliant new hires
Imagine hiring an exceptional analyst. On day one they’re talented but no use to you on a fund fact sheet. They don’t know which performance figures are final, how your compliance team reads the marketing rule, which disclosures apply in Luxembourg versus Ohio, or whose sign-off a client letter needs.
Over time the firm turns that talent into a productive colleague. It gives them system access, the house style, the approved data, a record of past decisions, and a clear sense of what they can do alone and what needs a second pair of eyes.
AI works the same way. Most firms have hired the talent. Few have built the onboarding.
Without a harness, every agent stays a day-one hire, and compliance has to supervise each one from scratch.
Why this matters now
Two things have shifted. Capable AI is no longer scarce, so the bottleneck has moved to the capacity to review and trust its output. And the market has moved from experimenting with prompts to investing in the environment around agents.
A harness can be bought or built. The knowledge inside it can only come from you.
Your fund data, rule interpretations, disclosures, brand and approval culture are the advantage. The harness is how you put them to work.
What it’s worth to the business
Measure success in compliant content reaching clients sooner, not in the number of agents deployed.
What we’ve seen in practice
Working with the marketing operations team at a leading global asset manager, we put two agents into production on a shared foundation of the firm’s own knowledge: one updating structured documents, one fixing PDF accessibility issues. A person approved every change. Updates ran 30–50% faster and accessibility fixes 40–60% faster.
The bigger win was trust. Every change came with a clear record of what was done and why, which made the compliance conversation easier.
Those were two agents, each carrying its own copy of the firm’s knowledge. The next step is a shared harness that every agent draws on, and that’s where the real test lies for marketing leaders in asset and wealth management. The firms that pull ahead will keep their fund data, rules and brand standards in one place rather than scattered across teams and tools. They will be able to show a regulator exactly what an agent knew when it created a piece of content. Each new agent will cost them less than the last, and their technology will be shaped around how the firm works rather than the firm bending to fit the technology. For most, the next investment won’t be another agent. It will be capturing what the firm knows, then building the harness to make the most of it.
A generic harness will give you the same agents as everyone else. A harness built around your context will give you agents no one can copy.
