How TalentGum Automated 95% of Customer Support with an AI Employee

The Problem

TalentGum teaches kids aged 5–14 real-world skills — keyboard, chess, dance, public speaking, coding, guitar, art, creative writing, and more. They run live classes with real teachers across 40+ countries.

Growth was never the problem. Conversations were.

Every parent who discovered TalentGum had questions. Which course fits my 7-year-old? How do demos work? Can I reschedule Saturday's class? Why can't my child log in? What happens if we miss a session?

These weren't complex problems. But every single one required a human to read the message, pull up the child's profile, check internal policies, and respond. At 62,000+ messages a month across WhatsApp, Instagram, and web chat, the support team was buried in structured, repetitive work.

Hiring more agents wasn't a scalable answer. The cost per conversation was rising, response times were slipping during peak hours, and parents in different time zones were waiting hours for simple answers.

"We were spending 80% of our support team's time answering questions that followed the same pattern every single time. The information existed — we just didn't have a way to deliver it instantly."

— Saunak Ghosh, Co-Founder & CEO, TalentGum

The Journey

Phase 1: Lead Qualification

Sagepilot started small, a lead qualification layer on WhatsApp. The AI greeted parents, collected the child's age, course interest, and location, then handed warm leads to the admissions team.

But parents didn't stop at sharing details. They kept asking: "Which course is best for a shy 8-year-old?" "How are the piano classes structured?" "Do you have group classes for chess?"

The AI was already answering. TalentGum decided to let it.

Phase 2: Full Pre-Sales Conversations

Sagepilot expanded into a complete enrollment assistant:

• Course recommendations based on age, interests, and learning goals across 10 subjects

• Pricing by region, automatically detecting whether a parent is in India, the Middle East, or the US and presenting the right plans

• End to End Demo booking - qualifying the parent, booking a demo and seamlessly connecting them to an enrollment specialist

• Curriculum deep-dives, explaining level structures (Beginner through Expert), session formats (1:1, 1:2, 1:5), and what each course covers

Parents got instant, accurate answers at 2 AM on a Tuesday. The sales team got pre-qualified leads with full context instead of cold inbound messages.

Phase 3: Post-Purchase Support (Where It Gets Interesting)

This is where Sagepilot became TalentGum's primary support infrastructure.

The AI employee — named Ananya — now handles the full lifecycle of a paying customer:

Identity & Context Awareness

• Automatically identifies the parent from their WhatsApp number

• Pulls up their child's profile, active subscriptions, enrolled courses, teachers, and upcoming sessions

• Handles families with multiple children enrolled across different subjects

Class Management

• Cancellations — Validates timing rules (45-minute policy window), processes cancellations with automatic credit-back, and handles edge cases like classes already in progress

• Rescheduling — Fetches available slots for 1:1 sessions, presents options, validates the new time, and executes the reschedule

• Hold & Resume — Places subscriptions on pause with defined start/resume dates, warns about teacher reassignment for long holds, and resumes when the family is ready

• Permanent Schedule Changes — Transfers subscriptions to a different teacher or batch timing while maintaining enrollment continuity

Real Example:

Parent (11:47 PM IST): "Hi, please cancel tomorrow's keyboard class for Aarav. We have a family function."

Ananya (11:47 PM IST): "Hi Meera! I've cancelled Aarav's Beginner Keyboard session with Teacher Priya, scheduled for Saturday, March 22nd at 10:00 AM IST. The session credit has been carried forward to your account."

Total time: 4 seconds. No human involved.

Live Class Support

• Shares joining links for upcoming and live sessions

• Troubleshoots login issues with step-by-step guidance

• Escalates persistent technical problems with full context

Policy Enforcement (Without Sounding Like a Robot)

• Enforces the 45-minute cancellation window — but if a parent is frustrated, escalates to the team for a case-by-case review instead of repeating the policy

• Handles regional pricing rules (EMI options for India, multi-duration plans for international)

• Manages course-specific constraints (coding is 1:1 only, no staggered slots for group classes)

Intelligent Escalation

• When something genuinely needs a human — refund disputes, repeated quality concerns, complex scheduling conflicts — Ananya creates a detailed support ticket with the full conversation context, student details, and issue summary

• The human team receives a complete brief, not a vague "customer is upset" handoff

• Parents are told "our team will reach out within 24 hours" — and they do, because the ticket has everything needed to act

Timezone & Holiday Awareness

• Converts class times for international parents automatically

• Adjusts messaging during holidays (like Eid or Diwali breaks) with updated schedules and modified response expectations

What Changed

Before Sagepilot

• 20-person support team handling repetitive queries

• 15-45 minute average response time during business hours

• No coverage outside IST business hours

• Parents in the US, Middle East, and Southeast Asia waiting overnight for simple answers

• Support agents spending 80% of time on structured, repeatable tasks

• Inconsistent policy application across agents

After Sagepilot

• 1-2 agents handling only complex escalations

• Under 10-second response time, 24/7/365

• Full coverage across all time zones

• Consistent policy enforcement with empathetic edge-case handling

• Support team focused on high-value interactions: parent counseling, quality reviews, retention

Why It Works

Most AI chatbots answer FAQs. Sagepilot's AI employee actually does things.

The difference is operational depth. Ananya doesn't just tell a parent "you can cancel on the dashboard." She checks if the class is within the cancellation window, validates the session status, processes the cancellation, confirms the credit, and tells the parent exactly what happened — in one message.

This matters because parents don't want information. They want resolution. And resolution requires the AI to understand policies, access live data, make decisions, and execute actions — not just retrieve text from a knowledge base.

"The AI doesn't feel like a chatbot. Parents tell us it feels like texting someone on the team. That's when we knew it was working."

— Saunak Ghosh, Co-Founder & CEO, TalentGum

Key Capabilities Deployed

Looking Ahead

TalentGum is expanding into new markets and adding courses. With Sagepilot, scaling support is no longer a hiring problem — it's a configuration update.

The AI employee that started as a lead qualifier now runs the majority of customer conversations. Not because it replaced the team, but because it handles the volume so the team can focus on what humans do best: building relationships with parents and improving the learning experience for kids.

For TalentGum, customer support is no longer limited by team size. For Sagepilot, this is what scalable AI infrastructure for education looks like.