Analysts stopped assembling the analysis and started reading it
Collecting data, reading financial documents, building the report and preparing the analysis came before any decision could be made. The judgement was the valuable part, and it was the last thing anyone got to.
The challenge
The work that took the hours was assembly: pulling numbers together, reading statements, chasing context, formatting it into something a person could reason about. Recurring analysis and routine reporting consumed the same effort every cycle, which meant the analysts were paid for judgement and spent their time on preparation.
- Financial data
- Analysis agent
- Research agent
- Risk model
- Executive report
What we implemented
- Financial analysis agent
- Analyses the data and surfaces trends, anomalies and the changes that matter.
- Financial statement agent
- Reads income statements, balance sheets and cash-flow statements and produces a first analysis.
- Research agent
- Gathers the market, company and economic context an analyst would otherwise collect by hand.
- Investment memo agent
- Turns the data and the research into a structured investment analysis to work from.
- Anomaly detection model
- Flags unusual transactions, movements and departures from the expected pattern.
- Forecasting models
- Revenue, cash flow, expenses and the other metrics that get projected every cycle.
- Risk analysis agent
- Evaluates the defined risk factors and marks what needs a human to look at it.
- Executive reporting agent
- Turns raw financial data into the concise report management actually reads.
The result
Analysts spend less time collecting and preparing, and more time interpreting and deciding. Recurring analysis and reporting that used to take hours is produced in minutes, with a person reviewing rather than assembling.