The problem: a call list nobody could finish
Around 200 parties a week needed a follow-up call. That is a full-time job on its own, and it was being absorbed by a team that also had ledgers to close, disputes to reconcile and an office to run.
The predictable thing happened: the list got worked from the top down. The largest accounts were called, the long tail of smaller parties was not, and the accounts that quietly slipped into 150 and 200 day buckets were usually the ones nobody had spoken to in a month.
What changed: the whole list, every cycle
The outstanding data now flows into PraecisAI, which scores every party into one of four stages: Soft Reminder, Follow-up, Strong Follow-up and Escalation. The day-ranges for those stages are set to match how the business actually trades, not a generic 30-day template.
PraecisAI then calls every party on the list in the configured time slots, in natural Hindi and English, and sends WhatsApp reminders carrying a branded PDF statement. The tone shifts with the stage: a polite nudge early, a direct and firm message once an account is deep into escalation.
Crucially, the calls are not independent. Each conversation carries the history and summary of the previous ones, so a party who promised payment on a date is asked about that specific promise rather than being asked to start the story again.
The result
The week of calling that a person could not finish is now finished overnight. In Rajesh Mehta’s words, recovery improved 40% in the first month.
The second-order effect matters as much as the number: the team stopped spending its day on routine reminder calls and started spending it on the accounts that genuinely needed a human, which are the disputes and the escalations the AI transfers across live.
“We were manually calling 200 parties a week. PraecisAI does it overnight. Recovery improved 40% in month one.”