User guide
GradWorkbench is where you run a graduate application campaign: the universities you're considering, the professors at each one, what you actually know about their work, how well it fits yours, the emails you send them, and every deadline that follows.
It replaces the spreadsheet, the browser tabs, and the sent-mail folder you'd otherwise be holding all of this in.
What it does for you#
- Keeps your list in one place. Every university and professor, with the department, the lab, the links, and where each detail came from.
- Researches professors for you, a few at a time. Their recent work, their group, whether they're taking students. Every finding is checked against its source — by a second assistant that is refused if it tries to confirm its own work — and waits for you to publish it before it reaches the record.
- Keeps what you know but would never say. Private notes on any record steer the research and the drafts, and are never quoted in anything a professor reads.
- Scores fit against your background. Not a vague "good match" — a score per dimension and specific, evidence-backed ways your work connects to theirs.
- Drafts your outreach. A formal and an informal version for each professor, written from your profile and the angle you picked. You edit and send them yourself.
- Tracks everything with a date on it. Follow-ups, replies, interviews, application deadlines, offer decisions.
Two things it deliberately does not do: it never sends an email for you, and it never treats an unverified finding as fact.
Start here#
| # | Chapter | What you'll do |
|---|---|---|
| 1 | Getting started | Create your account, fill in your profile, learn the layout |
| 2 | Building your list | Import a spreadsheet or add universities by hand |
| 3 | Researching professors | Run research batches and review what comes back |
| 4 | Reading a match | Understand the fit score and pick your angle |
| 5 | Outreach | Draft, edit, approve, send, and follow up |
| 6 | Applications | Checklists, interviews, offers, decisions |
| 7 | Dashboard and analytics | Read your numbers and find what's behind them |
| 8 | Settings | Profile, documents, email rules, research limits, security |
| 9 | Questions and fixes | Common questions and what to do when something looks wrong |
| 10 | Connecting your assistant | Set up the assistant that does the research — one page per tool |
| 11 | Every screen, part by part | What each control on each screen is, with annotated screenshots |
If you have half an hour and want the shortest useful path, read chapter 1, do the guided setup, then jump to chapter 3.
Research runs in your own AI assistant, not in this app. Chapter 10 is how you connect one, and nothing in chapter 3 happens until you have. It takes about five minutes, and there is a page for each assistant — Claude Code, Claude Desktop, Claude on the web, Codex, Cursor, Windsurf, Antigravity, VS Code, OpenCode, or anything else.
How the whole thing fits together#
Work left to right. Nothing downstream is available until the step before it exists — you can't match a professor nobody has researched, and you can't draft from a match that doesn't exist yet. The dashboard always tells you which step you're waiting on.
A note on your data#
Your workspace is yours. Uploaded documents, your profile, and your research stay in it, and nothing is shared with other workspaces. No email leaves the app: drafts are written for you to copy, send, and then log.