Do we need a technical team to use this?
No. You describe your procedure in plain terms — the steps and the scoring scale — and the dashboard is built for instructors and program directors: timelines, scorecards, a review queue, and plain-language explanations on every number.
Will this replace our expert reviewers?
No — it multiplies them. ActionQuality proposes; your experts validate. The least-trusted results surface first, every score links to its moment on video, and a human decision always wins over a machine one. That division of labor keeps your experts in charge — and removes the wait.
How do we know the scores are any good?
Because we grade ourselves the way you'd grade a trainee: on material the system never saw during setup. Every figure ships with its uncertainty, comparisons mark differences that could be noise, and small samples are labeled as small samples. When a number is shaky, the dashboard says so — out loud.
What kinds of procedures does it handle?
Anything with a repeatable sequence of steps a camera can see. Today that spans surgery and surgical simulation, judged sports routines, industrial assembly, fitness workouts, and instrument practice — and adding a new procedure means writing a definition, not commissioning a new product.
What happens when video alone isn't enough?
We tell you. Our favorite example is piano: the system can read a piece's difficulty from video, but a player's finesse lives largely in the sound — so we publish that limit instead of shipping a confidently wrong score. Knowing where the edge is, is part of what you're buying.
Can instructors correct the system?
Yes, right in the report — step boundaries and scores alike. Corrections are checked against the procedure, kept in history, and credited to the human who made them. Your experts' judgment becomes part of the permanent record, not a sticky note on top of it.
How fast is feedback?
In a published study, an expert panel took 21 days to turn around 50 trainee videos. With ActionQuality, a recording's timeline and scorecard are ready as soon as processing finishes, so reviewers start from evidence the same day the video is captured.
Where does our video go — and who can see it?
Your recordings stay yours. For pilots we process video on EU-hosted infrastructure, with on-premises processing available for operating-room footage. Recordings are used only to assess your sessions — never to train models for anyone else — and are deleted on a schedule you set. Clinical footage should be de-identified before upload (no patient identifiers in frame or filename), and we sign a GDPR data-processing agreement before any footage moves. The surgical numbers on this page were measured on HeiChole, a public research dataset — not on customer video. This website itself runs no analytics and sets no cookies.
Is this a medical device?
No. ActionQuality reviews recordings after the fact to assess the operator's skill — it plays no role during a procedure and makes no claims about patient outcomes or care decisions. It sits in the same category as an instructor reviewing tape, not clinical software.
Can it score OSATS or GOALS automatically?
The scorecard is rubric-agnostic — OSATS, GOALS, or your program's own scale plug into the same product, and every automated score is shown beside the instructor's mark with a human review queue on top. Our surgical validation so far is phase timelines on the public HeiChole dataset (81% six-fold cross-validated frame accuracy); what is and isn't yet validated is published in the honesty ledger.
Does it work with our rubric?
Yes — the scorecard adapts to whatever scale your field uses. A 1-to-5 clinical rating, a 0-to-110 judging score, and a 0-to-9 grading ladder all run through the same product, unchanged.
