# University

> A plan for a university or a teaching lab: the Professor's assistant plans courses and tutors students, and the Scientist backs the research.

StellarFirm for universities puts two AI Agent Assistants first: the **Professor's assistant** and the **Scientist**. The Professor's assistant prepares your courses and tutors your students. The Scientist backs the research behind them. You are the CEO: the instructor or the head of the lab. You approve every email and calendar invite that leaves the chat.

## Who this is for

| | |
| --- | --- |
| Company | A university department, a school, a teaching team, or a teaching and research lab |
| Team size | 1 to 50 |
| Main goal | Spend less time preparing material and more time with students and research |
| Start with | One Professor's assistant for your main course, then a Scientist for the research group |
| Biggest risk | A student handing in work the assistant wrote, or an unchecked solution reaching a class |

## Your assistants

| Assistant | Role in this plan |
| --- | --- |
| [Professor's assistant](/docs/personas/teacher) | Course syllabi, lesson plans, worked examples, exercises with checked solutions, quizzes and exams, rubrics and feedback, slides and handouts in LaTeX, and one-to-one tutoring |
| [Scientist](/docs/personas/scientist) | Literature searches with checked sources, experiment plans, symbolic and numerical checks, reproducible analyses, and papers in LaTeX |

Hire a Professor's assistant per subject if you teach more than one, for example one for programming and one for mathematics, each with its own focus.

## Suggested house rules

```text
House rules
1. The Professor's assistant gives hints before answers. Full solutions only when I ask for them.
2. The Professor's assistant never writes graded work for a student to hand in.
3. Every solution, example, and computation is checked before it is shown.
4. The Scientist never invents a result, a source, or a citation, and never calls a conjecture proved.
5. Emails to students and calendar invites wait for my Approve.
6. Course policies and grades come from me. The assistants never make them up.
```

> [!IMPORTANT]
> These rules are already how the Professor's assistant and the Scientist work. Writing them in your company brief makes them visible to everyone on your team. See [house rules](/docs/ceo/house-rules).

## Preflight

- [ ] Sign in at https://stellarfirm.ai/app with your invitation.
- [ ] Connect your mail, calendar, and files on [Integrations](https://stellarfirm.ai/app/integrations): [Gmail](/docs/integrations/gmail) or [Outlook](/docs/integrations/outlook), [Google Calendar](/docs/integrations/google-calendar), and [Google Drive](/docs/integrations/google-drive), [OneDrive](/docs/integrations/onedrive), or [SharePoint](/docs/integrations/sharepoint).
- [ ] Put one folder per course in your drive: the syllabus, past exams, and reading lists.
- [ ] Hire a Professor's assistant for each subject you teach, and a Scientist for your research group.
- [ ] Write your house rules in the company brief.

## Launch: Day 1

1. **Plan one lesson.** Start with the next class you teach.

```prompt title="Plan a lesson"
Professor's assistant, plan a [length] lesson on [topic] for [course and level]. Give learning outcomes, the order of activities, one worked example, and a short check for understanding at the end.
```

2. **Write exercises with checked solutions.**

```prompt title="Exercises with solutions"
Professor's assistant, write [number] exercises on [topic] for [course], from easy to hard. Check every solution before you show it, and give a hint for each one.
```

3. **Start the research brief.**

```prompt title="Research brief"
Scientist, write a research brief on [question]: what is known, the key papers with their sources, the open problems, and what we could test first.
```

4. **Approve what goes out.** Anything sent to students waits in [Approvals](https://stellarfirm.ai/app/approvals).

## First week

```prompt title="Course syllabus"
Professor's assistant, draft a syllabus for [course]: learning outcomes, a week by week plan, readings, and how students are assessed. Leave the grading policy for me to fill in.
```

```prompt title="Quiz and rubric"
Professor's assistant, write a [length] quiz on [topics] with a marking rubric and model answers. Check every answer.
```

```prompt title="Slides in LaTeX"
Professor's assistant, make lecture slides in LaTeX for [topic], with one worked example per idea. Publish them so I can edit them beside the chat.
```

```prompt title="Tutor a student"
Professor's assistant, tutor me as if I were a student in [course] who is stuck on [problem]. Ask me questions and give one hint at a time. Do not give the full answer.
```

```prompt title="Literature search"
Scientist, find the most cited and the most recent work on [topic]. Give each source with a link, one line on what it shows, and say which you could not check.
```

> [!NOTE]
> **Checkpoint.** By Friday you have a lesson, an exercise set, and a quiz you would use, and a research brief your group has read.

## Steady orbit

- **Every week.** Ask the Professor's assistant for next week's lesson and exercises, and read them before class.
- **After every assignment.** Ask for feedback drafts against your rubric, then edit them yourself.
- **Every month.** Ask the Scientist for a literature update on your group's topics.
- **Every term.** Ask the Professor's assistant to update the syllabus from what worked.

You can set the weekly preparation as a [routine](/docs/ceo/routines-and-loops), so it is waiting for you every Monday.

```prompt title="From research to teaching"
Scientist, summarize our latest result on [topic] in plain words. Professor's assistant, turn that summary into a short lesson for [course].
```

## The rest of the team

The Professor's assistant and the Scientist lead here, and the other assistants are there when you need them:

- **Coder:** works with you on a course site, a small app for a lab exercise, or to turn an analysis into a tool.
- **Copywriter:** helps you edit grant text, announcements, and papers in your own words.

## Pitfalls

- **Answers instead of hints.** Keep "hints before answers" in your house rules, and ask for full solutions only for your own answer keys.
- **Unchecked material.** Read every exam and answer key before it goes to a class.
- **Made-up policies.** Give the Professor's assistant your real course policies and grading scheme. It never invents one.
- **One Professor's assistant for everything.** Hire a Professor's assistant per subject so each one keeps its own focus.

---

Source: https://stellarfirm.ai/docs/playbooks/university