TL;DR

Baillie Gifford, the Edinburgh manager overseeing around £200bn, is a founding member of a new AI partner programme from behavioural analytics firm Essentia Analytics. The offering at its centre, Essentia Data-Out, delivers analysis of how well a firm’s portfolio managers make decisions directly into the systems where that firm runs its data and AI. The idea is that a manager weighing a trade could ask an internal AI tool how it compares with their past patterns.

What the product does

Essentia has spent more than a decade measuring skill and behaviour among equity portfolio managers, using a methodology it calls Behavioral Alpha and describes as peer-reviewed. Data-Out ships it into a client’s cloud environment through a Snowflake data share, where the firm’s models, agents and oversight systems can draw on it.

Essentia’s pitch is that asset managers feed their AI systems plenty of market and company data but almost nothing about the quality of their own judgement. Chief executive Clare Flynn Levy put it as: “Your AI is only as good as the data it can consume.”

As a founding partner and existing client, Baillie Gifford gets early access to each release and a say in which metrics Essentia builds next.

Why a long-term manager wants it

James Taylor, Baillie Gifford’s head of investment analytics, said the firm uses outside partners to bring in distinctive data alongside its own analysis, so that investors can understand their decisions on an evidence basis. The stated goal is to “systematically learn from the decisions we make”.

That is a different use of AI from the one most financial services coverage focuses on. Rather than generating recommendations, the system is being asked to hold up a mirror to human decisions, checking a manager’s reasoning against their own track record. It sits closer to the concerns in our analysis of adviser deskilling, where the risk is people deferring to AI instead of sharpening their judgement.

Looking forward

The announcement comes from Essentia, and neither firm has given results or costs, so this is a statement of intent rather than evidence. For other UK fund managers, the interesting question is governance: behavioural data about named individuals flowing into AI agents raises questions about who can query it and how it is used in performance reviews. Watch whether other founding partners are named, and whether Baillie Gifford reports anything on how its managers actually use the tool.