A tutoring platform that builds a profile of each learner, turns a school’s existing coursework into adaptive practice, and shows teachers where a class is getting stuck — without adding to anyone’s marking pile.
Classrooms still run on a one-speed operating system
Thirty students, one pace. The learners who need another pass get the same lesson as the ones who finished early, and the teacher only finds out who was lost when the test comes back. Most “AI tutors” answer that with generic content that ignores what the school actually teaches — and with data practices a district can’t sign off on.
The brief I set myself: adapt to the individual student, work from the school’s own material, give teachers visibility instead of homework, and treat student data as the school’s property from day one.
Four steps from sign-up to a class that is actually being tracked
Learner profile
A short diagnostic establishes each student’s starting point and how they take information in. Everything after is pitched to that profile.
Your own coursework
Teachers upload the PDFs, worksheets and reading they already use. Zaltiv turns them into lessons, practice and revision instead of generic filler.
Adaptive practice
Questions get harder or easier based on performance; time goes to the topics that aren’t secure yet.
Teach-Back Mode
Students explain a concept in their own words and get follow-up questions — surfacing shallow understanding a multiple-choice quiz misses.
Spaced revision
Material returns at widening intervals, weighted toward what each student keeps getting wrong.
Knowledge Map
A per-student and per-class view of mastered vs. unsecured topics, so intervention happens before the test.
Around the core sit the pieces a school needs before rollout: institution admin, bulk onboarding by class or year group, a guardian portal, study rooms for collaborative work, and a Security Center and Compliance Vault that state plainly what the software does with data.
Learning science in the product, security in the perimeter
The adaptive engine draws on documented effects from cognitive science — the pre-testing effect, interleaved practice, spaced retrieval, incubation — each shipped as a named feature (the Synaptic Half-Life Tracker, the Socratic Saboteur, the Circadian Governor) rather than a vague “AI” label.
Data lives in managed Postgres with row-level security on every table; sign-in runs through Supabase Auth with a separate super-admin role; retention changes, exports and purges are written to an append-only audit log. The Citadel, Zaltiv’s security operations center, moves beyond passive database logging to active perimeter monitoring — built by someone who has run IT operations and knows a platform is only as strong as its edge.
What isn’t in place yet is stated on the site rather than implied: verifiable parental consent, a formal DPA, an independent security assessment and a VPAT are on the roadmap, and districts are invited to ask for a current compliance summary before procurement.
Free tier live, pilots next
Zaltiv is free to start with no card required, and a demo can be booked for a department or whole school. The next milestones are classroom pilots, the compliance items above, and a published implementation guide for districts.
You don’t run next-generation hardware on an obsolete operating system. Why are we doing it to students?