Introduction
A licensed Australian financial advice firm of around 30 staff was already paying for an AI assistant, and most staff had tried it. Daily use was limited, no written rules governed it, and leadership could not confirm it was the right platform to build on. The firm engaged CoterieLabs to resolve the platform question and design its AI operating model.
The challenge
Nearly every licensed user had tried the assistant, but use was confined to chat and a few prompts a day. Almost all of the firm’s automation sat with one staff member. Metered charges rose with each new automation, so the cost of scaling was unknown.
There was no firm-wide policy on what staff could put into AI or who signs off its output. As a licensed firm holding client financial data, the platform choice had to be defensible.
Our solution
CoterieLabs defined five weighted evaluation criteria with the firm, then assessed three platforms on realistic advice tasks using synthetic client files, never real client data. Outputs were scored blind, including by two of the firm’s own staff.
Each provider’s data and contract terms were checked against 43 dated source documents, and current usage was measured through read-only access to the firm’s own admin console. The AI use policy was written in plain language first; the chosen platform was then configured, setting by setting, to enforce it.
The impact
Most advice firms already have AI tools. What they lack is a decision they can defend and rules their staff can follow. By agreeing the criteria first, testing on the firm’s own type of work and writing the policy before rollout, the firm replaced scattered experimentation with one platform, one rulebook and a staged launch, with evidence behind each step.
Figures are from the engagement’s own evaluation. Client name, platforms and vendors withheld.
The result
Before
An AI assistant most staff had tried but few used in depth, automation that sat with one person, a metered cost nobody could forecast, and no firm-wide policy on AI use.
After
- One platform approved for all staff by senior management, 24 days after signing
- On a test client file with 12 issues to find, the chosen platform found all 12; the incumbent, as staff used it, found 8.5
- Preferred by the firm’s own reviewers in 9 of 14 blind comparisons
- A fixed, capped annual cost for all staff, known before rollout
- AI use policy issued to the pilot group within five weeks of signing
- Every platform setting recorded against the policy section it serves, with rollout in three waves, pilot first
If your firm pays for AI it can’t yet defend
Book a 30-minute call. We’ll look at the tools your staff already use, the rules you’d need, and what a platform decision would rest on.