Campus Platform: AI Teacher Agents
Overview
A college group with several departments wanted prospective and current students to get department-specific answers online, at any hour, without every question landing in a faculty inbox. Algon built a set of AI teacher agents, one per department, each presented as a member of that department's faculty and answering in its voice. The agents sit inside the wider campus platform, alongside the department pages they draw on.
AI agent per department
Student access
Full platform build
Responsive
The challenge
we had to solve
Student questions are highly department-specific: a query about lab schedules in one stream has a different answer, and a different tone, from the same question in another. A single general-purpose chatbot would blur those boundaries and, left ungrounded, would invent course details. The institution also needed to know what students were actually asking, and to be sure that anything sensitive or out of scope reached a person rather than a guess.
How we
approached it
Each department's agent runs on a large language model constrained by a department-specific prompt template and persona, and answers from that department's own syllabus, FAQ and page content via retrieval-augmented generation: content is chunked, embedded and stored in a vector index, and only the most relevant passages are placed into the model's context window for each question. Guardrails keep answers within the department's scope, decline when retrieval finds nothing reliable, and route admissions, fee or personal queries to staff through a human hand-off. Every conversation is logged so faculty can review answer quality and spot gaps in published content.
Agent per Department
A separate assistant for each department, with its own persona, tone and scope rather than one generic bot.
Grounded Answers
Retrieval over embedded department content keeps responses tied to what the institution has actually published.
Guardrails & Hand-off
Scope checks, refusal on low-confidence retrieval and a clear route to human staff for sensitive questions.
Conversation Insight
Logged conversations show faculty which topics students ask about and where published content falls short.
Four phases
zero shortcuts
Department Knowledge Audit
Collected syllabus, FAQ and page content per department and defined what each agent may and may not answer.
Retrieval Pipeline
Chunked and embedded department content into a vector index, with retrieval scoped to each department.
Personas & Guardrails
Wrote prompt templates and faculty personas, then added scope checks, fallbacks and human hand-off.
Launch & Review Loop
Embedded the agents in the platform and set up conversation logging for ongoing faculty review.
The work,
in full resolution
Drag to explore. Click to open full screen.
Before vs.
after
The platform completely changed how students interact with us online. The AI agents and 360 view made a huge impact.
Want results like these? Let's talk about what we can build together.
Start a project