Building Arka's Auditable Scam-Call Decision Engine
Separated evidence extraction, deterministic scoring, and hard rules in a real-time fraud product.
- Applied AI
- Fraud Detection
- Decision Systems
- Founder Product
Ayan Aggarwal
Founder at Unimind
From their time as
Founder
Unimind • 2026 - Present
Overview
Ayan is building Arka, an AI-powered scam-call detection product that assesses fraud risk during live calls. The product uses an auditable hybrid decision system, with a consumer wedge designed to support a later BFSI offering.
The Story
Ayan started Arka under Unimind to help users recognize scam risk during live calls before they act. The product listens for evidence, assesses fraud risk in real time, and presents a warning to the user.
He set the decision engine around a strict separation of concerns. An LLM extracts evidence, deterministic code calculates the score, and hard rules can override the result, keeping decisions auditable and reducing dependence on changing model behavior.
Ayan is leading the product as a founder, with a working MVP and an OTP-block feature functioning end to end. He is also handling fundraising, incorporation, partnerships, distribution planning, hiring choices, and cash allocation.
The market path combines consumer and enterprise logic. Ayan sees the consumer product as a way to build distribution and a fraud corpus, while the enterprise path gives banks and fintech companies a clearer route to revenue.
His near-term goal is to raise funding and bring the Android product to an initial cohort of thousands of waitlisted or pilot users. These remain targets rather than achieved outcomes.
Ownership Snapshot
Broad role
Founder owning product, go-to-market, partnerships, and fundraising.
Goal
Build a consumer scam-call product and a path toward a BFSI offering.
Direct ownership
Set the decision architecture, product direction, and initial market model.
Team execution
Coordinates a lean build while developing investor, fintech, telecom, and OEM relationships.