Live

AI Course Tutor

Runs a 90-day AI course in chat: daily lessons, Q&A and graduation certificates

An AI tutor that delivers a structured daily lesson on Telegram or WhatsApp, answers students' questions in the course's own teaching style and Cantonese tone, and issues certificates at graduation. Students get a one-to-one coaching feel at class scale.

TelegramWhatsAppDaily lessonsStudent Q&ACertificatesAdmin dashboardWeekly review

All agents

At a glance

What it does01

Sends daily lesson cards on a fixed schedule: morning hook, teaching, practice prompts and a night recap video
Answers student questions any time, using the course's own knowledge pack and persona
Attaches PDF cheat sheets and recap videos to each day
Issues a graduation certificate, sends a congratulations video and collects testimonials
Gives the course owner an admin dashboard for students, progress and chat review

Example conversation02

AI Course Tutor

online

ClientHow should I change the Day 12 prompt so it sounds more like my personal brand?

AI Course TutorFollow the three steps from today's lesson: define your audience first, then add the tone you normally use, and finish with an example. Try rewriting it that way and paste it back here, and I'll take a look.

Illustrative example

Human vs agent03

Same taskRunning one week of a course cohort

By hand

≈ 10 hstaff time

Send each day's lesson to every student by hand
Answer about 60 questions from 30 students
Track who has kept up with each day

With the agent

≈ 1.5 hstaff time

Scheduler sends each day's lesson on time
AI answers questions in the course's own style
You review flagged answers once a week

Students get answers any time, day or night

By hand

With the agent

Saves ≈ 85% of staff time

Estimates for one typical task, based on how the workflow runs. Actual results vary.

Ad-hoc AI vs our agent workflow04

The same task, done by prompting a general chatbot by hand versus through the agent’s set workflow.

Prompting a chatbot by hand

Tokens per task

Right first time

Course notes pasted in again for every question
Answers drift from the course's teaching style
Nothing stops answers going beyond the syllabus

Our agent workflow

Tokens per task

63% fewer

Right first time

Router picks the right course pack for each question
Weekly review turns fixes into new rules
Regression check keeps correct answers correct

No AI is 100% reliable. When a run fails, the agent retries automatically or hands it to a person.

Estimates for one typical task, based on how the workflow runs. Actual results vary.

How it works05

Step 1
The course owner plans the day's lesson
Step 2
Lesson cards, a PDF and a recap video are prepared
Step 3
A scheduler delivers them to every student
Step 4
Students reply or ask questions in the chat
Step 5
The AI picks the right course pack and answers within scope
Step 6
A weekly human review turns fixes into new rules

Why it's trustworthy06

Answers stay within the course's own material and teaching style
Every week the owner reviews flagged answers, and fixes become new rules or knowledge entries
A regression check makes sure a change never breaks an answer that was already correct