$250K+
Additional ARR Generated
99%+
Information accuracy across 150K+ chats
25%
Reduction in ad costs via Conversions API
About the company
CoLearn
CoLearn is one of Southeast Asia's fastest-growing K-12 edtech platforms, on a mission to move Indonesia from the bottom 10% to the top 50% of the global PISA rankings. Founded by Abhay Saboo and Marc Irawan, the company offers live interactive STEM classes taught by top Indonesian educators for just $7/month, reaching over 4.8 million students since 2020. Over 80% of users report improved school grades, and the platform was recognized as a Google Play Best App and one of LinkedIn's Top Startups in Indonesia. CoLearn has raised $34M from Alpha Wave Incubation, GSV Ventures, Sequoia Capital India's Surge, AC Ventures, BINUS Group, and others.
The Challenge
CoLearn’s $7/month price point required massive scale, but their manual sales process created a bottleneck that conflicted with their unit economics.
01
Scaling beyond human limits
During peak season, volume surged to 3,000+ conversations/day (one every 29 seconds). Handling this manually would require 30 reps.
02
Unit economics of a $7 product
High-headcount sales teams are unsustainable for low-cost, high-volume subscription models. Capacity had to be decoupled from headcount.
03
High stakes for information accuracy
In education, incorrect details about schedules or pricing erode parent trust instantly. Accuracy is non-negotiable for brand survival.
8 full-time reps
Managing all WhatsApp enrollments
30+ minutes
Response time during peak volume
Previous failed AI attempt
High error rates and frequent downtime
$7/month
Unit price making human sales inefficient
What didn't work
Why Generic Chatbots Failed
CoLearn was early to AI, but their first vendor couldn't meet the rigorous demands of Indonesian parents.
01
Fabricated details (Hallucinations)
The bot quoted wrong pricing and fake schedules, leading to lost sales and eroded trust.
02
Unreliable uptime
The system crashed frequently, students clicking ads landed in dead-end conversations.
03
Rigid iteration cycles
When accuracy issues were flagged, the vendor took weeks to update information or fix errors.
04
Static knowledge base
Failed to stay current with shifting enrollment windows and seasonal pricing updates.
05
Friction-heavy handoffs
Bots only handled FAQs, forcing parents to wait hours for a human to actually close the sale.
Shelved. The team returned to fully manual operations before partnering with Covena.
What Covena does differently
The Covena Autonomous Agent
Covena built a sales agent that handles the full funnel—from first click to final payment.
01
End-to-end closing capability
The agent doesn't just "chat"; it qualifies, handles objections, and sends payment links to close the deal.
02
Meta Conversions API Integration
Payment completions are fed back to Meta's algorithm, optimizing ad spend for actual buyers, not just clickers.
03
99%+ Information Accuracy
Built around CoLearn's actual SOPs to ensure every schedule and price quoted is 100% verified.
04
Instant, Scalable Response
Handles 3,000+ daily conversations (1 every 29 seconds) with a sub-5 second response time.
05
Full Funnel Autonomy
The entire journey from ad click to paid subscription happens in one thread with zero human involvement.
Deployed in 6 weeks. 8 reps reduced to zero.
Building a Fully Autonomous Enrollment Channel
Covena replaced the 8-person sales team with a custom AI agent integrated directly into WhatsApp and CoLearn's marketing stack.
Instant Ad Engagement
Parents move from a Meta ad to a live enrollment chat in seconds with no forms or landing pages.
Natural Language Qualification
Fluently understands grade levels, subject needs, and intent in both Indonesian and English.
Autonomous Closing
Generates and sends payment links directly in the thread once the parent is ready to enroll.
Bilingual Mid-Thread Switching
Switches between Bahasa Indonesia and English seamlessly based on the user's preference.
Conversions API (CAPI)
Synchronizes payment data with Meta ads to lower CAC and improve lead quality.
Full AI BPO Model
A hybrid model where AI handles the frontline and a managed team handles rare escalations.
The AI Sales Agent in Action
See how the agent handles the journey from an initial ad click to a completed tuition payment on WhatsApp.
Implementation: Live in 6 Weeks
Weeks 1–2
Discovery & Workflow Mapping
Mapping enrollment SOPs, qualification criteria, and parent psychology.
Weeks 3–4
Build & Edge-Case Testing
Simulating hundreds of chats across grade levels and languages to ensure 99%+ accuracy.
Weeks 5–6
Rollout & Scaling
Gradual traffic transition from human reps to the AI agent until 100% volume.
Post-Launch
Optimization
Continuous refinement of the Conversions API to drive down ad costs.
The Impact of Autonomy
Moving from a human-dependent model to a scalable AI sales channel.
$250,000+
Additional ARR
Revenue generated directly by the AI agent with zero human involvement.
25%
Ad Cost Reduction
Improved efficiency through the Meta Conversions API integration.
150,000+
Conversations Handled
Massive scale handled without adding a single sales representative.
0
Sales Reps Needed
Successfully transitioned from an 8-person team to a fully automated channel.
Metric
Before
After
Sales Team
8 Reps
0 Reps-100%
Incremental ARR
-
$250,000+↑ $250K+ Growth
Response Time
Minutes (Higher at peak)
Seconds (Instant)↑ 95%+
Peak Daily Volume
Manual / Limited
3,000+ ConversationsUnlimited Scale
Information accuracy
Variable (Human error)
99%+↑ Significant
Ad Cost Efficiency
Baseline
25% reduction↑ 25%
Operating Model
8-person in-house team
AI-powered (AI + managed escalation)Scalable
Abhay Saboo
,
Co-founder & CEO, CoLearn
Scaling Education for Millions
With enrollment fully automated, CoLearn is focusing on curriculum excellence and student outcomes to move Indonesia up the global PISA rankings.
Ready to automate your sales channel?
See how Covena builds AI sales agents that handle the full funnel from first message to payment.
