01
Student Lookup by Phone Number
The chatbot resolves the student from the number they message from, so schedule and account questions are answered against their own record.
The team could answer every question a student asked; they just could not get to them fast enough. A WhatsApp agent now replies the moment a query arrives.
An EdTech company was struggling to keep up with the sheer number of support queries coming from students: schedules, platform problems, general course questions. The team knew the answers; there were simply more questions arriving than hours in the day, and students waited.
More questions arriving each day than the support team had hours to work through.
The team could answer everything a student asked. The problem was how long students waited in the queue before anyone got to them.
Schedule and platform queries repeated constantly, consuming the time needed for the questions that genuinely warranted a person.
Many questions could not be answered at all without first knowing which student was asking.
We built a WhatsApp-based AI chatbot that replies the instant a message arrives. It identifies the student from the number they are messaging from, answers from the company's own question bank, and escalates anything it is not confident about to a support agent.
The chatbot resolves the student from the number they message from, so schedule and account questions are answered against their own record.
Retrieval runs over the company's existing question bank, so general answers come from approved content rather than the model's own recall.
Where the chatbot is not confident, the conversation is handed to a support agent rather than guessed at, leaving the team the queries that genuinely need a person, and the time to answer them.
Schedule and deadline reminders delivered on the channel students already use, rather than a portal they have to remember to check.
Every question the chatbot struggled with is logged for review and addition to the question bank, so coverage improves with use.
Students get an answer as soon as they ask, instead of waiting in a queue.
Routine schedule, platform and course queries resolved without occupying the team.
Low-confidence questions go to an agent, and each one is logged for the question bank.