Balanced class rosters, built from what actually works
Class placement is one of the highest-leverage decisions a school makes each year — and one of the most manual. Gnosis IQ studies how your students have done and which teachers lift which learners, then places every student into the class most likely to raise their outcomes — balanced, fair, and the right size.
From spreadsheets to a finished roster
Four steps. Most of the work is a one-time upload — after that, a new roster is minutes away.
Upload your grade history
A few years of past grades — one row per student, per class, per year. Gnosis builds each student's learning profile and each teacher's track record from it. The more years, the sharper the recommendations.
Add your classes — or don't
Already know next year's classes and who teaches them? Upload them and Gnosis places students into the best-fit section. Don't have a class list yet? Gnosis builds balanced classes from your candidate teachers.
Generate the roster
A seeded optimization engine searches thousands of arrangements for the placements most likely to lift outcomes — while honoring periods, capacities, and class sizes.
Review, then publish
See the predicted lift over a round-robin baseline, drill into the rationale behind every placement, then export to CSV or a OneRoster bundle — or publish to your SIS.
Six things it balances at once
Good rostering is a balancing act. Auto-Roster optimizes every objective together, instead of trading one off for another by hand.
Predicted outcomes
Places students with the teachers most likely to lift their results, using each teacher's real track record — not guesswork.
Learning-profile mix
Spreads ability and achievement evenly across parallel sections, so no single class carries an unbalanced load.
Class sizes
Keeps enrollment even across sections — no class left overflowing while another sits half-empty.
Equity floor
Minimizes the number of students predicted to fall below a healthy outcome threshold — lifting the most vulnerable first.
Emotional load
When students use check-ins, balances emotional state — stable, watch, at-risk, volatile — across classes, automatically.
Hard rules respected
Never double-books a student's period, honors keep-apart pairs, and respects every section's capacity.
Private by design
Auto-Roster's algorithm runs entirely on Gnosis's own servers against your own data — never sent to any outside AI service or third party. It's a deterministic engine: the same inputs always produce the same roster, so every placement is explainable and auditable.
100% local algorithm
No large language models, no third-party AI. Your student data never leaves your district's environment.
Grounded in your own data
Recommendations come from your district's real history — not generic national benchmarks.
Explainable, every time
Each placement carries a rationale: the candidate sections considered and why one won.
OneRoster-ready
Export finished rosters as CSV or a OneRoster 1.1 bundle, or publish straight to your SIS.
Auto-Rostering questions
Still have one? Talk to our team.
What does Auto-Rostering need to get started?
One file is essential: a few years of grade history — one row per student, per class, per year. To place students you also provide your current students and teachers, unless you sync them from a live SIS. Next year's class list is optional; Gnosis can build balanced classes for you.
How does Auto-Roster decide where a student goes?
It learns each student's learning profile and each teacher's track record from your grade history, then an optimization engine searches for the set of placements that maximizes predicted outcomes while balancing class composition, size, and equity — and respecting hard scheduling rules like periods and capacity.
Does it use AI that sees our student data?
No. The rostering algorithm runs locally on your own data and is never sent to any outside AI service or third party. It is a deterministic, seeded algorithm — the same inputs always produce the same roster — so every placement is explainable and auditable, not a black-box guess.
What happens to a student with no grade history?
New students with no track record are seated by balance alone — counted toward class size and composition but excluded from outcome predictions, so nothing is fabricated.
Can we keep the classes we've already planned?
Yes. If you already know next year's classes and teachers, Auto-Roster places students into them. If you don't, it builds balanced classes from your candidate teachers — and you can switch approaches later.
How do we get the roster into our SIS?
Export the finished roster as a CSV or a OneRoster 1.1 bundle, or publish it to the API for your SIS to consume.
Ready to build next year's rosters in minutes?
See Auto-Roster run on a sample of your own data and watch the predicted lift for yourself.
Request a Demo