AI for education: every leading AI model, for the whole campus, under your control.
Faculty, staff and students on one account with ChatGPT, Claude, Gemini and dozens more, admin controls over who has what, private conversations, and work that is never used to train a model.
- One account, every model
- Admin controls per workspace
- Never trains a model
GeminiAutoClaudeChatGPTTrusted by professionals at
The campus problem
AI arrived on campus one personal account at a time
Faculty pay for their own tools. Students use whatever is free. The IT office cannot say which products hold what, and the policy meeting is about a landscape that changed again last month.
Today
With Magai
No inventory of tools
A department buys ChatGPT, a lab buys Claude, a professor expenses an image tool. Nobody can list what the institution is paying for.
One account for the institution
Every model in one place, one invoice, one list of who has a seat. The inventory is the admin screen.
No control over access
A personal account cannot be given to a class, restricted to a course, or switched off at the end of term.
Workspaces per course or department
A workspace for a course, a lab or an office, with its own members, files and context. Add students for the term, remove them after.
Data going to consumer products
Lesson plans, student work and research drafts pasted into tools whose default is to train on them.
Nothing trains a model
Prompts, files and outputs never train a model. That is true for every model on the account, so the policy is one sentence.
Every model is a new decision
A new model ships and the question of whether to allow it starts over, in a committee.
New models arrive inside the same rules
When a provider ships a new model it appears in the same workspace, under the same privacy terms, with the same controls. No new decision.
On your work
A term's work, on the right model each time
Teaching, research and administration each lean on a different strength, and all three sit in one thread.
Claude for careful writing
Lesson and assessment design
Learning objectives into a lesson plan, then a quiz, then three versions of the quiz at different reading levels.
ChatGPT for volume
Feedback at scale
An agent built on the rubric drafts first-pass comments on forty submissions; the instructor edits rather than starts.
Gemini for long documents
Research and literature
Summaries with sources, long PDFs read in full, and a running conversation about a paper that stays with the paper.
Auto picks the model
Accessibility
Alt text, plain-language versions, transcripts and translations of course material, in one pass.
DeepSeek for reasoning
Student use, within bounds
A course workspace with an agent that tutors rather than answers, and the instructor's rules baked into it.
Perplexity for sources
The office
Newsletters, policy drafts, event copy and the reply to the same parent email, drafted on a schedule.
Agents
The agents a department builds first
An agent is a set of instructions written once, by the person who knows the subject, that then runs on any model for anyone in the workspace.
@Rubric marker
Reads a submission against the rubric and drafts feedback in the instructor's tone.
@Socratic tutor
Asks the next question instead of giving the answer. Built for a course workspace.
@Plain-language editor
Rewrites material at a chosen reading level without losing the content.
@Literature scout
Summarizes a paper, lists its claims, and finds three related sources.
What the policy meeting asks
"Where does student and faculty work go?"
To the model provider, for that one reply. Never into a training set, and never into a product the institution does not control.
- Never used to train a modelNothing you write, upload or generate becomes training data, for any provider, on any plan.
- Access ends when the term doesSeats and workspace membership are managed by your admins. A student who has finished the course has finished with the workspace.
- One privacy answer for every modelBecause every model runs inside the same account, the terms do not change when the model does.
What people say after they switch
Unedited, from the review platforms.
Plans that fit a department or a campus
Start a department on Standard today. For a campus, enterprise sales has volume rates and an invoice procurement will recognize.
Standard
For individuals who are just getting started with AI tools.
$20/mo
- All Standard Models + Tools
- Unlimited Agents
- 100 Scheduled Task Runs
- Standard Customer Support
- Add Users $20/ea
Pro
For serious users who need more capacity for daily AI work.
$40/mo
- All Premium Models + Tools
- Unlimited Agents
- 1,000 Scheduled Task Runs
- Priority Customer Support
- Multi-account Integrations
- Add Users at Custom Rates
Ultra
For power users who need maximum monthly capacity.
$200/mo
- All Premium Models + Tools
- Unlimited Agents
- Unlimited Scheduled Task Runs
- Priority Customer Support
- Multi-account Integrations
- Add Users at Custom Rates
30-day 100% money-back guarantee on every plan. Costing it for a team? Use the calculator, or talk to enterprise above ten seats.
What schools and universities ask first
Magai is also made for
The same workspace, written to a different desk.
- EnterpriseEvery model for the whole organization, on one invoice
- Marketing AgenciesA workspace per client, one bill for the team
- ChurchesSermons to socials, for one small bill
- ConsultantsYour method as an agent, a workspace per engagement
- CoachesNotes, follow-ups and programs in your voice
- HospitalityGuest replies and reviews, on time, in your voice
- Health ProfessionalsThe paperwork side of the practice, in one private workspace
- LawyersDrafting and review, never used for training
- Social Media MarketersHooks, captions, images and clips in one thread
- Content CreatorsScripts to thumbnails, for one subscription
Give the whole campus every model, under one set of rules.
Start with one department this week. Bring the rest when the policy meeting has seen it work.