AI agents & agentic AI
Meet your AI agent
Most people have met a chatbot. An agent is different: it takes steps — opens your own Studio analytics, reads what is actually there, and only then answers. For a publisher that means asking, in plain words, how your courses are doing across every school that redeems them.
What is an AI agent, in plain words?
No jargon. Two short answers and one real example.
A chatbot talks. An agent takes steps.
A plain chatbot answers from whatever it already knows, in one go. An agent works like a colleague with your filing cabinet open: it works out what it needs, opens the right analytics page, reads the rows that are actually there, and only then writes an answer. The numbers come from your catalog, not from the model's memory.
“Agentic AI” is AI that acts within limits you set.
Agentic simply means the AI can act — use a tool, run a lookup, go back for one more detail — instead of only producing text. The limits matter as much as the ability: which tools it may use, whose data it may open, what it may do with them. In EduGears AI Studio those limits are the design. The agent can read your own courses, and reading is all it can do.
Example
What that looks like
How is my Circuits course doing across schools?
Circuits & Signals is live in 7 institutions, with 412 learners across 9 cohorts. Completion is 63% overall, but two cohorts sit under 30% — both redeemed the code in the last three weeks. Learners stall most often in Module 4, “Transient response”. Want the learner table for those two cohorts?
Where this came from
Course Analytics · 9 cohortsCog opened the same Course Analytics page you can open yourself, counted the rows, and linked it. Click the source and you land on that page, with the same filters applied.
In your workspace
How the assistant works in EduGears AI Studio
Cog is already there — the sparkle in the corner, on every page. Ask in plain words, and it answers from the analytics your account already has.
Ask across every school at once
One question covers every institution that redeemed your access code: enrolments, completion, cohorts labelled per LMS. You don't pick a report, set a filter or export a spreadsheet — you ask, and Cog opens what it needs.
Find where learners get stuck
“Which module loses people?” “Who redeemed the code and never started?” Cog reads the cohort and learner tables behind your course, answers with the rows it found, and offers the next cut.
Follow a code's reach
Ask which institutions are running a course, when they redeemed it and how far their learners have got. It is the redemptions report you already have, in a sentence.
Answers you can open
Every figure arrives with the page it was computed from. Click through and you are on the real analytics page — same course, same filters — so you can check the number or keep digging by hand.
Guardrails, not promises
The limits are built into how the agent runs — not merely written into its instructions.
Read-only, always
Cog can look; it cannot change anything. No course edited, no access code regenerated or revoked, no learner record touched, no message sent. Reading is the only thing the agent is able to do.
Your own catalog, nothing wider
It answers only about courses your organization owns, as the person asking. An institution's private data stays theirs: you see the learners taking your course, their instructor still sees only their own section — the same privacy line your analytics already draws.
Every number is traceable
Answers name the page they came from, and the source opens it with the same filters. Figures are computed by the same analytics your team reads, never by a free-form database query — and where the data isn't there, Cog says so instead of guessing.
Your switch, in Settings
Course-data questions are a capability your organization's admins turn on or off for everyone, at any time, under Settings → AI & Usage in EduGears AI Studio. Turning it off leaves Cog answering everything else.
Questions publishers ask
What is an AI agent?
An AI agent is an AI that takes steps to answer you instead of replying from memory alone. It works out what it needs, uses a tool to look it up — in EduGears AI Studio, one of your own analytics pages — reads the result, and then writes an answer based on what it found.
Can it change my courses or my access codes?
No. Cog is read-only by design and by construction: it has no tool that writes, edits, publishes, regenerates a code or deletes anything, and the lookups behind its answers run on read-only connections as a backstop. Editing a course or managing a code stays your action, in the normal screens.
Can it see an institution's private data?
No. Cog answers about courses your own organization owns, and only about the learners taking those courses — the same cross-institution analytics your team can already open. Everything else inside a school's LMS stays with the school, and their instructor still sees only their own section.
Where do the numbers come from?
From the same Studio analytics your team reads: cohort completion, learner tables, redemptions. Cog uses those pages' own calculations rather than inventing a query, which is why its figures match the page — and every answer links back to it, so you can check.
Do we have to set anything up?
No. Cog is part of EduGears AI Studio with nothing to install — open the assistant from the sparkle in the corner and ask. Organization admins can switch course-data questions on or off under Settings → AI & Usage, and AI is included in your plan, with fair usage.
Ask about your own courses
Cog is already part of EduGears AI Studio. Open the assistant and ask how a course is doing across the schools that run it — or start free and publish your first one.