Loading your page…
Loading your page…
Knowledge Gap Detection
Suture audits your second brain for you — flagging notes that are outdated, untested, isolated, or on topics you keep asking Osler about — so you can close the gap while it’s still cheap to close.
Priority-ordered — outdated first. Every flag says exactly why it was raised.
The problem
Gaps don’t announce themselves. They hide inside notes that look complete, feel familiar and have never actually been tested — until an exam or a patient asks the question.
Research on physician self-assessment shows it often disagrees with observed performance — how covered a topic feels is not evidence that it is.
A note citing a guideline version that has since changed looks exactly like a current one. Nothing about the page says “stale”.
A note you have never quizzed yourself on is a claim of knowledge, not proof of it — and orphan notes with no links fade from view entirely.
Four signals
Suture watches for four gap signals, priority-ordered: outdated — a note cites a guideline version that has since changed; revisited — you keep asking Osler about a topic you already have a note on; untested — notes you’ve never quizzed yourself on; and isolated — notes with no links to anything else. Each flag tells you exactly why it was raised.
“Heart failure — GDMT overview” cites ESC Heart Failure 2021. A newer version has been published — the note doesn’t know that yet.
The cross-app tell
The strongest tell comes from across the ecosystem: when you keep asking Osler about a topic you already wrote a note on, that note clearly didn’t stick. Suture connects the two — a pattern of gold Osler questions becomes a violet flag on the exact note that needs work.
When do I start anticoagulation in new AF?
Monday
CHA₂DS₂-VASc — when is a DOAC indicated?
Wednesday
Today
Atrial fibrillation — anticoagulation
Last edited 5 months ago
Quiet and private
Three of the four signals are pure database logic — fast, private, and free of AI cost. The revisited check runs a cheap similarity gate before any AI confirms a match, and the whole audit runs off the chat path. Your questions get answered at full speed; the gap report assembles quietly in the background.
Your chat
“First-line treatment for status epilepticus?”
Database checks
No AIOutdated, untested and isolated — pure queries
Similarity gate
GatedA cheap filter runs before any AI is involved
AI confirms
Last stepOnly revisited matches that pass the gate
Three of four signals never touch AI — computed from your own notes and activity.
From gap to fixed
A gap report is only useful if it ends in action. From any flagged note, jump straight in to update it, quiz yourself on it to turn untested into tested, or link it to related notes so it stops being an island. Each fix makes your second brain more complete than it was.
Nephrotic syndrome
Flagged by your gap report
Quiz complete — 8/10 on this note just now
Feeling prepared and being prepared are different states. Four priority-ordered signals show you which one you're in — note by note, before the exam runs the same audit for you.
Notes on trust
With Gap Detection
Four priority-ordered signals audit your second brain continuously — the same system that catches an untested note before finals catches a stale one in practice.
The audit runs quietly in the background — your only job is the fix, and every fix leaves your second brain stronger than the flag found it.
Suture scans your notes for the four gap signals — quietly.
Each flagged note shows exactly why it was flagged.
Update the note, quiz yourself on it, or link it to related notes.
Every closed gap makes your second brain more complete.
Why an external audit beats a feeling of preparedness — the research behind gap detection.
Let your second brain audit itself — four signals, clear reasons, and a next step for every gap it finds.
Free forever for students · No credit card required