Simulation
Your customers, simulated.
Not a survey and not a spreadsheet. Your actual customers, as agents, living through the months ahead, once as things are and once with your decision.
How it works
Six parts, one simulation.
Agents fitted to your customers
Every customer becomes an agent with their own habits, learnt from their history: how often they buy and how much, when a subscriber is likely to cancel, when a contract is likely to renew. Customers who bring friends bring them in the simulation too.
An AI panel for what's never happened
Your history can't say how customers react to a price you've never charged. So a panel of AI personas, drawn from your own customer groups, is asked. Their answers adjust the agents, corrected by changes you really made in the past, and informed by the reports you upload and confirm.
The world outside your business
Seasons learnt from your own year, public holidays where you sell, the weather where you trade, fuel prices, exchange rates and any monthly series you upload. Each is kept only if it makes the forecast better. Competitor pages are watched for you.
Forecasts and scenarios
Thousands of runs, month by month, with the range they land in, by group. A scenario runs the same customers with the same luck and one thing changed, so the difference you see is your decision. The cost in AI credits is shown before anything runs.
Deep mode
A small cast of your customers lives through the months one by one, each deciding in their own words. Slower, and the place to find out why, not just how much.
From result to action
Turn a forecast into a plan in one step. Avir watches the months as they come in and raises a task when reality leaves the range.
End to end
From your data to a decision, in an afternoon.
- 01
Connect
Point Avir at your orders, subscriptions or invoices. Read access is enough to start.
Minutes - 02
Understand
Avir finds your customer groups and fits each customer's habits, then tests itself on your recent months.
Minutes - 03
Forecast
Twelve months ahead, by group, with the range they'll likely land in.
Seconds - 04
Try a decision
Describe the change in plain words. The AI panel answers for each group, and the agents run again.
About a minute - 05
Act
Make it a plan, watch it against actuals, and get a task when it drifts.
One click