Recap & Recording: XLcompass – Treasury Automation That Works

Key Takeaways

Freddie Turner

Start with repeatable treasury tasks

Reconciliation, transaction tagging, cash sweeps and reporting are strong starting points for automation. Freddie said: “We solve the initial obvious problems first of reconciliation, transaction tagging for cash flow forecasting, and so on.”

Let forecasts learn from variances

AI can review forecast differences and use the findings in the next cycle. As Freddie explained: “It will observe any differences, so every day it runs, it will look at a variance, understand why it was.”

Ben Hipwell

Automate the path to a liquidity decision

One client combined transaction categorisation, reconciliation and liquidity recommendations. The result was a much earlier start to daily decision-making: “The team moved from taking their first big decision at late morning or lunchtime to taking their first big decision at 8am when they first come in.”

Keep AI explainable and auditable

AI recommendations need to be supported by evidence that remains available later. Ben put it simply: “You need to be able to explain why AI got to a certain answer.”

Martin Scheid

Forecasts need an explanation

Treasury needs to understand why cash flow changes, not only what the forecast says. Martin summed up the requirement as: “Tell me why.”

Use AI within clear guardrails

AI can help interpret data and policies, but treasury teams still need control over decisions. Martin said: “They want to get the work done, but they want to decide by their own.”