HelloPM · Assignment Examples Browse 149 examples →
Meaningful AI Improvement · Zerodha

AI Recovery Agent for Retail Investors Who Have Lost Money

An in-app recovery agent for futures and options traders who have taken a big loss and are about to chase it

Meaningful AI Improvement FinTech Zerodha
Zerodha logo

The problem they went after

Retail investors trading futures and options hit large drawdowns and then revenge trade, which compounds the damage. Investment apps offer nothing structured for breaking that cycle. There is no personalised recovery pathway, no behavioural coaching, and no mechanism that notices when someone's trading pattern has tipped into dangerous over-leverage or panic-driven decisions. The user is left alone at the exact moment judgement is worst.

What they actually did

The Recovery Agent works in three layers. A loss detection engine flags high drawdowns alongside behavioural signals like rapid leverage spikes, and triggers a Recovery Mode onboarding flow. Personalised recovery plans then lay out SIP-based roadmaps with equity and debt allocation suggestions and timeline simulations a user can activate in one tap. Behavioural coaching closes the loop through an AI assistant that explains past mistakes, sets pre-trade guardrails and sends weekly progress reports.

What you can take from it

Worth stealing

  • Detect, intervene, coach as a three-layer structure
  • Leverage spikes used as a behavioural trigger signal
  • One-tap SIP roadmap out of Recovery Mode
  • Success defined as fewer repeat losses, not more trades

Read this if

PMs whose most active users are the ones getting hurt

The transferable bitDetect, intervene, coach is a structure that ports to any product where a user's own behaviour is what harms them. The other transferable choice is how success is defined: fewer repeat losses in futures and options and a shift toward sustainable behaviour, which means the feature wins by reducing activity rather than by increasing it.

The submission itself

Open the PDF →

Reading on a phone? The embedded viewer is cramped — open the PDF in a new tab instead.

Everything else they shared

+Same industry, different takeFinTech