A LITTLE CONTEXT
A learner’s question often belongs to a particular lesson. SelfHelp AI keeps the conversation beside that material, inside the GoHighLevel portal.
The chat is one part of the product. Behind it is a workspace for the people running the course: a place to manage content, see what members are asking, and help when an answer needs their attention. I built the experience around those two connected jobs.
Find the useful moment.
Answers draw on course content and lesson context. Transcript timestamps can send someone straight to the relevant part of a video, so finding an answer also brings them back to the learning.
The course workspace brings together the syllabus, lesson coverage, transcripts, documents, and common questions. It makes the material available to the support team as well as the learner. Course and page settings give the assistant a useful starting point for each part of the portal.
Keep the team in the conversation.
Questions that need a person become tickets. The team can respond through the Inbox, Telegram, or Google Chat, with the reply delivered in the member’s portal conversation.
The Inbox gives that handoff a clear place to happen. A teammate can review the conversation and its context, work through the open question, and follow the reply back to the member. Support stays connected to the learning experience.
Turn a useful reply into a better answer.
A team can review and publish a useful answer for reuse. MCP and REST access let other agents consult that knowledge too.
The knowledge workspace separates courses, notes, Q&A, open questions, and documents. Search, tags, and pinned notes help the team find the material they need, while a Team Assistant can use that shared context. The result is a product that supports the people answering questions as well as the people asking them.
What interests me here is the connection between assistance and authorship: AI helps people find information, while the team decides what becomes trusted material.
Build the application around the assistant.
SelfHelp AI uses Next.js and React for the interface, PostgreSQL and Prisma for application data, and Qdrant for knowledge retrieval. The system connects course content, conversations, member progress, and the team’s support workflow.
This is the kind of AI product I enjoy making: the model has a useful role, and the surrounding software gives people the controls, context, and everyday tools to work with it.
Help one learner. Make the next answer easier.
WHAT’S UNDERNEATH
Tools behind the experience.
- GoHighLevel
- Next.js
- React
- PostgreSQL
- Prisma
- Qdrant
- OpenAI
- MCP
- REST
- Telegram integration