Customer story
An edtech platform resolves 74% of support tickets with AI.
A Delhi EdTech platform handled 2,400 support tickets a month, three-quarters of them repeat questions answered from scratch. Webzenia built an AI chatbot grounded in the course database to resolve them instantly.
The challenge
A Delhi EdTech platform for competitive exams was handling 2,400 support tickets a month, and 74% of them were repeat questions. Students asked the same things about schedules, access, and course content, and a human answered each one from scratch.
The support cost scaled with enrolment instead of staying flat. Every Rs. 48 spent resolving a ticket bought an answer the team had already written dozens of times. The volume left agents no room for the questions that genuinely needed a person.
The approach
Webzenia built an AI support chatbot wired directly into the platform's course database. The bot answered from the live catalogue, so a student asking about a specific batch got the real schedule, not a generic reply.
We used LangChain to ground the model against the database and OpenAI and Anthropic for the responses, which kept answers accurate to the platform's own content. The retrieval layer meant the bot could not invent details that were not in the course data.
Deflection logic ran on n8n. The bot resolved what it could and escalated the rest into Zoho with full conversation context, so a human picked up where the bot left off rather than starting over.
We also exposed the bot on WhatsApp Business API, the channel most exam aspirants already used. The build ran 10 weeks, including evaluation against real ticket transcripts.
The results
The chatbot now handles 74% of all support tickets without a human, exactly the share that had been repeat questions. Cost per interaction fell from Rs. 48 to Rs. 9.
Student satisfaction held at 4.1 out of 5 even as automation rose. Answers came instantly and stayed accurate, because the bot reads from the live course database rather than a static script.
The result holds as the catalogue changes. New batches and schedules flow into the same database the bot answers from, so it stays correct without a rewrite.
The stack behind it
More customer stories
Your turn
Want results like these?
Tell us where you want to grow. We'll map the first 90 days and show you what's realistic for your market.
What happens next
- 1A 30-minute discovery callWe learn your goals and constraints
- 2A tailored growth planChannels, milestones, and a realistic timeline
- 3KickoffA senior team, not a junior handoff
Start the conversation