04 CREATED FOR ELISA
The Quiet Trial
An AI retention agent
An AI agent prepares each retention batch. People make the call.
WORKING PROTOTYPE / SYNTHETIC DATA
WHAT WE SAW
A familiar challenge.
A new possibility.
Retention teams bring together lists, signals and offers before they can act. The prototype explores how an AI agent could prepare that work while keeping the decisions with the team.
Sanna and Aava are fictional personas.
What we made
Aava, a fictional AI agent, identifies at-risk segments, prepares ranked options and explains what each recommendation is based on. Sanna, a fictional team member, approves, edits or rejects the proposal. Nothing moves without her yes.
Today
- Lists built by hand.
- One save offer for everyone.
- No learning from what worked.
With The Quiet Trial
- Segments identified and prepared by AI.
- Ranked options, each explained.
- Results inform the next cycle, with people in control.
THE MOMENT
A ready queue, one row per segment, with a person deciding the edge cases.
- Bring the at-risk segments into one prepared queue.
- Review ranked options and their supporting signals.
- Let the agent prepare the proposed next action.
- Approve, edit or reject each recommendation.
- Capture the outcome to inform the next cycle.
SEE IT WORK
The prototype, in action.
Illustrative prototype using synthetic data. Screen values demonstrate the flow.
A CLOSER LOOK
Small details. A different experience.
MAKING IT REAL
Built around
Elisa’s world.
The starting point
Designed around Pega, Marketing Cloud and Elisa’s existing churn and propensity models. A pilot could start with Finnish single-product households.
What we’d validate together
Understand the team’s current preparation workflow, test the quality of the agent’s proposals, and agree clear approval controls.
The screens and flows are illustrative. Customers, accounts and activity are simulated; the prototypes would be validated with Elisa’s teams.