The Intelligence Layer for Education.
Run your institution. Understand your students. Help your teachers act earlier.
- Multi-tenant
- Organization, branch, year
- Deterministic
- Scores never come from an LLM
- Auditable
- Every privileged action recorded
- Exam-aware
- JEE Main, Advanced, NEET
The problem
Most institute software records what happened.
Marks get entered. Attendance gets ticked. Fees get receipted. At the end of the term, everyone can describe the past in detail and nobody can say which student was slipping in September, which topic the whole batch never absorbed, or whether last month’s extra class actually helped.
The data was always there. It was just never connected.
What Wiser answers instead
- 1What happened?
- 2Why did it happen?
- 3What should happen next?
- 4Who should act on it?
- 5Did the intervention work?
The loop
One closed loop, not a folder of disconnected modules.
Every feature exists to keep this cycle turning. A number that nobody can act on, and an action nobody measures, are both dead weight.
- 01
Collect
Attempts, attendance, fees, syllabus — one schema.
- 02
Understand
Deterministic analytics down to the concept.
- 03
Predict
Mastery and risk signals, with confidence attached.
- 04
Recommend
A next best action, and the reason for it.
- 05
Act
A teacher assigns an intervention with an owner.
- 06
Measure
Did the intervention move the number?
Four portals
The same data foundation, seen four different ways.
Students
A single view of where they stand and what to do next.
- Score, accuracy, percentile and rank per test
- Drill-down from subject to chapter to concept
- Mistake patterns separated from knowledge gaps
- A study plan that reacts to the last assessment
Teachers
Their batches, their subjects, and nothing they should not see.
- Batch analytics scoped to what they actually teach
- Test and DPP authoring with a blueprint engine
- Weak-topic detection before the next assessment
- Interventions with an owner, a date and an outcome
Administrators
The institution as one picture, across every branch.
- Academic, faculty, finance and admissions in one place
- Filter by branch, year, batch, subject or teacher
- Collection rate, outstanding and overdue at a glance
- A full audit trail of every privileged action
Parents
Their own account, limited to their own children.
- Performance, attendance and fees without a student login
- Weekly summary of what changed and why
- Teacher remarks and upcoming assessments
- Payment history and receipts
Question intelligence
A question bank that knows what it contains.
Every question carries its exam, its place in the curriculum down to the concept, its provenance, and its licence terms. Difficulty is not asserted — it is measured from how students actually performed on it.
Once a question has been used in a graded assessment it is frozen. Corrections create a new version, so a three-year-old attempt still renders exactly as it was sat.
Ingest, structure, review
An adapter-driven pipeline: extract, normalise LaTeX and images, deduplicate, AI-tag, quality-check, then human review before anything is published.
Provenance and rights
Source, publisher, licence and commercial-use flags are recorded per question. Nothing is published without a source, and commercial use defaults to off.
Exposure tracking
Wiser knows which questions a student has already seen, so a practice set can exclude them — or deliberately revisit them for revision.
How every AI answer is produced
- DatabaseThe raw rows, under the same RLS as everything else.
- Deterministic analyticsScores, ranks and averages, computed in code.
- Structured contextOnly the numbers the user is allowed to see.
- Language modelExplains and summarises. Never calculates.
- Validated outputChecked against a schema before it is shown.
AI, kept in its lane
The model explains the numbers. It never invents them.
Ranks, percentiles, attendance percentages and fee totals are computed deterministically and stored. An LLM is never asked to do arithmetic, and never writes to an analytics table.
- Provider-agnostic, with fallback and structured outputs
- Prompt versions recorded against every request
- Token and cost tracking with a per-organization ceiling
- Assistants query a controlled analytics layer, never raw SQL
Analytics
Drill from a cohort to a single concept without changing tools.
Mastery, with confidence
A mastery score is always paired with how much evidence backs it, so nobody is judged on three questions.
Mistake intelligence
A careless slip and a conceptual gap need different responses. Wiser separates them, and a teacher can override the classification.
Time intelligence
Slow and wrong is a different problem from fast and wrong. Both are tracked per question and per topic.
Learning gain
Normalised gain across a period, so improvement is visible independently of where a student started.
Risk signals
Attendance decline, score decline and engagement drop combined into a level a teacher can act on.
Teacher effectiveness
A configurable model built on normalised learning gain — an input to administrative review, never an automated verdict.
Security
Isolation you can point at, not just promise.
Student records, assessment data and financial history sit in the same system. That only works if the boundaries are structural.
Row Level Security
Isolation is enforced in the database, not just the application. Every table has RLS enabled, and the policies are tested — cross-tenant, cross-branch, teacher scope and parent scope each have assertions that must pass before a migration ships.
Capability-based authorization
Code asks whether a user can do a thing, never whether they hold a job title. Role escalation is blocked at the database level: an admin cannot mint a role at or above their own rank.
Answer keys stay server-side
Students hold no read policy on the question table at all. Question content reaches them through one function that withholds the key and the solution until results are published.
Append-only audit trail
Score changes, fee changes, publishing and role changes are recorded with actor, before and after. No role has an update or delete policy on the audit log.
See it against your own institute’s data.
We will walk through the student, teacher, admin and parent portals, and show exactly how tenant isolation is enforced.