Researching professors

A spreadsheet row tells you a professor exists. It doesn't tell you what they're working on this year, whether their lab is active, or whether they're taking students. Research fills that in, and it does it in batches — a few professors at a time, deliberately.

First, connect an assistant. Research runs in an AI assistant you already have — not inside this app — and nothing in this chapter happens until one is connected. That is chapter 10, which has a page of exact steps for each one. It takes about five minutes.

3.1 Why it works in batches#

Researching a professor properly takes real time and produces a lot of claims that all have to be checked. Batches keep that honest and keep it paced:

• A few at a time, so you review findings while they're fresh instead of facing a wall of hundreds.
• Rotating between universities, so you don't end up with one institution fully researched and thirty untouched. Each batch takes at most one professor per university, then picks up where it left off next time.
• A weekly ceiling, so a campaign is a habit rather than one exhausting weekend.

The rotation position is remembered between batches, so coverage stays even across your whole list without you tracking whose turn it is.

3.2 Set your limits once#

Go to Settings → Research Batches before your first run.

Full screen

How much do you review?#

Findings are always checked against their sources first — that part isn't optional and can't be turned off. This setting only decides whether you sign off before they appear on the professor's record:

• Everything, before it publishes — safest, and the right choice for your first few batches. You see each professor's findings before they become part of the record.
• Only conflicts and warnings — the practical middle. You're pulled in when sources disagree or something couldn't be confirmed.
• Nothing, publish once checked — fastest. Use it once you've seen enough batches to trust what comes back.

If you pick either of the first two, findings wait for you in Review — and nothing downstream can use them until you publish. That is §3.4a, and it is the step people miss.

3.2a Who gets researched#

Two lists of tick boxes in the same settings page decide which kinds of people are worth spending research on: one for the kind of appointment (tenure-track, adjunct, emeritus, postdoc and so on) and one for seniority (assistant, associate, professor, distinguished chair). Both have to say yes.

Ticked to begin with: tenure-track, research professor, teaching professor, clinical professor, professor of practice, and Not established yet. Everything else is off.

These are defaults, not filters#

Any individual professor can be switched on or off from the Professors list, and that choice beats the settings. The Research column shows where each person stands:

To change it, open the row's menu and pick Edit status. If you later want the settings to take over again, set it back to Follow rule — that is not the same as Skip, and it is why the three options exist rather than a switch.

The Research filter above the list answers the resulting question — Will be researched or Skipped — after the settings and any override have both been applied. It is about what happens next, not about what has already been done; for that, filter on Stage.

Nobody is skipped before they have been looked at#

This is the part worth understanding, because it is deliberately not what you would guess.

A professor nobody has researched yet is always researched, whatever their appointment says. The rules only start deciding once an agent has actually read their faculty page.

The reason is that imported appointments are unreliable. When a roster comes in from a spreadsheet, the appointment is whatever the university's staff directory happened to say — and genuine tenure-track faculty regularly arrive labelled affiliate or adjunct. Five of the strongest matches in this workspace did. If those labels were trusted, those people would be dropped before anyone looked at them, and nothing on screen would show that it had happened.

Seniority is worse still: it is only ever worked out by research reading a faculty page, so every professor who arrives by import is Not established yet.

The practical consequence: on your first batch, these settings will not exclude anybody. Everyone is unresearched, so everyone is eligible. The tick boxes start mattering on later passes, once appointments are established. If you want somebody left out before then, set them to Skip individually.

3.3 Run a batch#

Full screen

1. Go to Batches.
2. Look at Next batch covers. It lists the exact professors this run would pick up, with their universities. Nothing has started yet.
3. If the selection looks right, click Run this batch.
4. You land on the batch's page, where each professor appears as a row you can follow.

The three cards at the top tell you where you stand: batches used this week against your limit, how many professors the next batch would cover, and how many batches you've run in total. Underneath, Batch history is every previous run, newest first — click any row to reopen it.

What a batch page shows you#

Each row is one professor, and its status is the story of that professor's research:

Most of these pass by in seconds and you will never catch them. The four worth recognising are Waiting on your review (it is waiting for you, and nothing downstream moves until you act), Failed the source check (research was done and then rejected — the record is unchanged, which is the system working), Failed — can retry and Failed, which look alike and are not: only the first is worth a second attempt.

Retry re-runs a professor that failed recoverably. Skip drops one out of this batch without losing it — it comes back around in a future run.

3.4 Why "checked against sources" matters#

Every fact recorded about a professor keeps the page it came from, and gets re-checked by a separate pass before it is treated as true. Nothing skips that step.

That's why the professor's Data badge matters, and why fit scores and email drafts only become available once a record is verified. The whole point is that when a draft says "your 2025 paper on X", that paper exists and says what the draft claims it says.

3.4a Publish what passed#

With review turned on, checked findings do not go straight onto a professor's record. They collect in Review → Ready to publish, grouped by professor, and wait.

Every part of this screen is explained piece by piece in Review, part by part — including what each badge means and why publishing is a sign-off rather than a judgement.

Open a row to see exactly what would go on the record — every fact, with the value that was found, and every publication. Then Publish, or Reject if something looks wrong.

Two things worth knowing:

• Nothing here is suspect. Everything in this queue already passed its source check, run by an assistant that was refused when it tried to confirm its own work. You are deciding when it goes live, not whether the research is any good.
• The queue blocks more than it looks like. A match can only cite a published publication. If research has run and a professor's page still looks thin, or a match refuses to attach the paper it is clearly about, this queue is why.

Rejecting is not destructive. The finding stays checked but unpublished, and re-running verification brings it back.

3.4b The other tab: possible duplicates#

Review has a second tab, Possible duplicates, and it answers a different question from the duplicate-skipping that happens during import. Import compares what you are loading against what is already there and skips exact repeats. This queue catches the ones that are not exact: the same person entered twice under two spellings, or a professor who appears in two of your sheets under different departments.

Each flag shows the two records side by side, labelled Newly imported and Already on file, with the department and the sheet row each came from, so you can tell which copy is the one you meant to keep. Needs a decision and Settled separate what is still open from what you have already dealt with.

Resolving a flag deletes one of the two records and marks the flag settled — one action, not two, because resolving without deleting leaves the duplicate in place and deleting without resolving leaves a flag pointing at nothing.

The app refuses to delete a record that has matches, drafts or publications attached to it, and tells you how many. That is a guardrail worth understanding rather than working around: if the copy you were about to delete is the one that has been researched and written about, deleting it would take all of that with it. Delete the other copy instead, or move the work across first.

Two near-identical professors are worth resolving before you research either of them. Research one copy and the other stays thin forever, splitting the evidence for one person across two records that neither matching nor drafting will ever join back up.

3.5 Read what came back#

Open any professor from Professors or straight from the batch page.

Full screen

Six tabs, in the order you'll use them:

• Overview — department, lab or group, email, faculty and personal pages, Google Scholar, ORCID, whether they appear to be taking students, and when the record was last checked.
• Research — the individual findings. Each one carries its status and the sources behind it, labelled by how much weight they carry: an official source (their own faculty or lab page), a reliable source (a reputable index), or a weaker one. Click any source to read it yourself.
• Publications — recent papers, each with the venue, what the paper contributes, and why it's relevant to you.
• Match — covered in chapter 4.
• Email — covered in chapter 5.
• Outreach — covered in chapter 5.

Two habits worth forming while reading a record:

• Open one source per professor. Not all of them — just enough to keep your trust calibrated.
• Read anything marked "Sources disagree" or "Couldn't confirm" before you write. These are the details most likely to embarrass you in a cold email.

3.6 When research goes stale#

Faculty move, labs close, funding ends. A record that was verified a long time ago is marked Needs refresh, and the dashboard raises it as upkeep rather than an emergency. Find them all under Professors → Stale research, and put the ones you actually care about back through a batch.

What's next#

You have verified research. Now find out how well it fits your work: reading a match →