• Result of the engagement
  • Services

Twenty-four employee ideas, six tests and two uses in daily work

A services company employing 40 people invited employees to submit repetitive tasks they would like to improve. Within two weeks, 18 people put forward 24 proposals. Each one described the current work, how often the task occurs and an example of the result needed.

Problem

The company had plenty of AI ideas but none tested in practice. The employees knew the repetitive tasks, not the management.

The change

In two weeks 18 people submitted 24 proposals. The person leading the AI area, together with the managers, assessed data availability and the ability to check quality, and picked six ideas for testing.

Effect

After eight weeks two uses entered regular work, two stayed in testing, and two were stopped because the effort of checking cancelled out the benefit.

Key result
Total working time on the two live uses fell from 52 to 28 hours over a four-week period. That recovered 24 hours, after accounting for checking and corrections.

Measurement

MetricBeforeAfterChange
Ideas submitted24 from 18 people
Selected for testing6
In regular work after eight weeks2
Working time over a four-week period52 h28 h−24 h

Basis of measurement: A four-week measurement before and after the change for the two live uses, at comparable task volume, including checking and corrections.

  • Industry – services
  • Team – 40 people
  • Submissions – 24 ideas from 18 people
  • Measurement period – 4 + 4 weeks

What we changed

The person leading the AI area, together with the managers, assessed data availability, the ability to check quality and the employee’s readiness to take part in a test. Six ideas were selected for testing. The remaining 18 were given a recorded reason for being deferred, for example missing data, low task frequency or dependence on another system.

After eight weeks two uses entered regular work. They covered preparing internal summaries and drafting first versions of answers to recurring client questions. Two more stayed in testing, and two were stopped because the effort of checking and correcting cancelled out the benefit.

Measurement and results

For the two live uses a measurement was carried out covering four weeks before and four weeks after the change. At comparable task volume, the total working time of everyone involved fell from 52 to 28 hours over the four-week period. Twenty-four hours of work were recovered, after accounting for checking and corrections. The team spent that time on current client matters.

What the owner gets

The owner saw the whole course of the initiative: submissions, selection, tests, the uses that went live and the decisions to stop. Twenty-four ideas were not presented as 24 deployments. The benefit was evidenced by the results of two uses that are actually in work, and every other idea had a clearly recorded status.

Conclusion
Twenty-four ideas were not presented as 24 deployments. The benefit was evidenced by the results of two uses that are actually in work, and every other idea had a clearly recorded status.

See all case studies

Your company

Let's talk about the first AI use case in your company.

Book a call