Earlier career
A financial operation that automated 95% of its settlement files with machine learning
35% fewer people needed, and 100% accuracy within four months.
Led by Amir Nair before he founded Amanstra. This was not an Amanstra engagement.
In numbers
35%fewer people needed
95%handled automatically
100%accuracy in four months
1,000+file formats received
The situation
Standing settlement instruction data reached a financial organisation in more than 1,000 file formats. Handling it took a lot of time and people, and poor data quality limited how much of it could be used.
What we did
- Applied machine learning to read the files.
- Eliminated, consolidated and standardised formats where possible.
- Taught the model 35 file formats, then tested it with randomised checks.
- Changed the content of the recognised formats, and introduced new ones.
The result
- 35% saving in people, without loss of data quality.
- 95% of files handled automatically, and 5% routed to people.
- 100% accuracy within four months.
More on this theme: Insights, or what we do.
You built the business. We help it scale.
Start a conversation