Agree.
Currently, it's a prolonged iteration phase and experimenation rather than set in stone practices.
The path to maturity will be prolonged.
- Todor
------------------------------
Todor Kostov
Director
------------------------------
Original Message:
Sent: 23-03-2026 07:47
From: Carlos Salas
Subject: Update on AI Adoption on the Trading Desk (FactSet)
Thanks Todor. This one is great since it offers a more detailed analysis focused on the "trading desk" business case rather than the usual aggregate economy AI reports.
From my experience, the "AI as a side tool vs a workflow capability" pitfall aka "Experimentation hell" is probably the most common issue nowadays with many applications stuck in this stage.
------------------------------
Carlos Salas
Portfolio Manager & Freelance Investment Research Consultant
Original Message:
Sent: 22-03-2026 19:09
From: Todor Kostov
Subject: Update on AI Adoption on the Trading Desk (FactSet)
Brief update by FactSet this week on the AI Adoption on the Trading Desk:
link
Where AI Is Making an Impact Today:
Reduce rekeying across pre-trade preparation, execution support, and post-trade follow-through.
Enable teams to sort what needs attention now versus what can wait, especially when workflows generate a high volume of small tasks.
Filter low-signal alerts and surface the subset that is more likely to change a decision.
Key Success Factors in AI Adoption:
1. Governance is also important with AI, particularly in these areas for trading desks:
Permissions and access rules should govern both the raw data and the insights generated.
Teams should be able to trace what was generated, when, and from what inputs so that AI-influenced decisions can be reviewed.
Firms should define low- versus high-risk tiers up front to apply the right controls while maintaining efficiency with AI.
If the system cannot answer reliably, it should say so.
2. Concrete metrics enable trading desks to measure the success of AI adoption. For example, teams might consider:
How much time AI removes from finding, validating, and contextualizing information.
Reduction of steps, handoffs, or time-consuming tasks.
Whether usage becomes habitual rather than occasional.
Improvement in consistency, avoidable errors, or executional/operational costs.
- Todor
------------------------------
Todor Kostov
Director
------------------------------