Keerthi Karan

Enterprise AI agents, from pre-sales to production

The model is the easy part.

I get AI agents working inside banks, insurers and carmakers. The hard part is the decision layer around the model, and I build it with the people who used to do the job by hand.

Bengaluru · Canadian citizen · Open to relocation

Deployed at

  • HDFC Life
  • Jana Small Finance Bank
  • Maruti Suzuki
  • Waaree Energies
  • Casagrand
  • Pilgrim
  • TVS Credit
Keerthi Karan
OYO · 2015
0 to 400+
corporate accounts, on a B2B vertical I opened
Amazon Canada · 2021
700+
people led as an L6 operations manager
Gupshup · 2023
$12M+
ARR portfolio in voice AI and CPaaS
ConvoZen.ai · 2025
24
enterprise AI deployments I own

What I believe about this work

Start with the business, not the model.

The first question is which decision a person makes a thousand times a day, and what a wrong one costs. The technology question comes later. It's usually smaller.

Write the rules down yourself.

Specs, disposition logic, the data cleanup. If a rule can't be written down, the agent can't follow it, and nobody notices until it's live. Some of what I've built.

The second deployment is the real test.

Go-live proves the technology. The next use case gets earned in the months after, through reviews, fixes and numbers people believe.

Case studies

Inbound support on voice AI

HDFC LifeLife insurance

I built the inbound agent that answers HDFC Life support calls, one of the first of its kind in Indian BFSI, and grew it from 1,575 calls a month to 16,000+. Plus renewal calling that cut ninety call outcomes to two questions.

41%of inbound support calls resolved with no human. 23% in September
Read the case study ›
Running AI at bank scale

Jana Small Finance BankBanking

50+ voice agents and 70+ dashboards for one bank, in four languages, at 10M+ calls a quarter. The AI collection agents now beat the human ones.

42% vs 35%collection rate, AI agents against human agents
Read the case study ›
Taking the human out of the loop

Waaree EnergiesSolar energy

Voice agents that read live stock and price mid-call, place the order, and chase the installer when a lead sits untouched.

0people between CRM lead and paid order. Before, every order waited on a callback
Read the case study ›
Winning the first yes

Maruti SuzukiAutomotive

A dealership receptionist, built as a POC and presented on stage at the Maruti Suzuki Accelerator.

4,000+dealerships in the follow-on plan for leadership
Read the case study ›
Keeping the agent's knowledge true

CasagrandReal estate

The inventory lived in a price sheet that changed every week. I turned it into data the agent can trust, so it never pitches a sold flat.

~90Kdials in its first two weeks live in production. Voice AI Pioneers award
Read the case study ›
Making the business case

Karur Vysya BankBanking

A cost model for pre-due collections and call-quality monitoring with voice AI.

₹5.8 vs ₹11cost per minute, AI agent against human calling

Get in touch

Email is the quickest way to reach me. I reply within a day.

keerthi.karan92@gmail.com