Platform Specification

The science behind
finding every child.

A complete view of the Kratu platform — its founding principles, evidence base, clinical safeguards, and the international frameworks it rests on.

This is a early detection-only platform and not diagnostic, with human oversight at every step — maintaining rigorous ethical standards. All early detection outputs require qualified clinician review before any referral, intervention, or communication is released.
Built for Trust

A platform engineered for clinical rigour, ethical AI use, and real-world classroom adoption.

Kratu is designed from the ground up as a early detection, validation, intervention and progress monitoring platform — never a diagnostic instrument. Every architectural decision flows from this principle: structured human judgement at every consequential step, with AI used to scale observation and personalise support, never to label children.

The platform's five-stage pipeline mirrors how real schools and clinicians work: teachers see every child, structured early detection triages concern, clinicians validate independently, and adaptive intervention is delivered with progress tracked across the school year. Each tier produces evidence that informs the next — and every output flows back into the platform's governed learning loop.

The result is a system that is transparent in its reasoning, conservative in its claims, and uncompromising in its safeguards.

Core Design Principles
No diagnostic labels. Outputs are framed as risk bands (low / monitor / referral suggested), never as conditions.
Strongest-language-first. Children are assessed in the language they know best, always.
Human-in-the-loop governance. The AI never sends a referral, generates a report, or contacts a family without clinician release.
Impairment gate. Symptom flags require concurrent functional impairment evidence before escalation.
Auditability by design. Every override, every change of judgement, every release is logged.
Multi-jurisdiction privacy. Compliance configured to applicable law per deployment region.

Built on four foundations

Every decision flows from these design commitments

Early Detection only
Identifies concern. Triages support. Never diagnoses.
Clinician-led
Qualified specialists hold final authority at every stage.
Evidence-grounded
Anchored to international frameworks and expert clinical review.
Ethically governed
Privacy, audit, and oversight built in — not added on.
The Science

Evidence-based from the ground up

Every item, threshold, and clinical decision point is grounded in established frameworks, Indian clinical guidance, and expert review — not retrofitted AI.

The platform is designed to be defensible to procurement committees, professional bodies, and parents — because what supports a child should be testable, transparent, and grounded in something other than a vendor claim.

01
Multi-Tiered System of Supports (MTSS)
Universal early detection at Tier 1, targeted group support at Tier 2, intensive individual intervention at Tier 3 — the internationally recognised graduated response framework.
02
Response-to-Intervention (RTI)
Progress monitoring data drives all escalation decisions. Strong, partial, minimal, and regression response patterns trigger pre-specified actions with clinician review thresholds.
03
DSM-5 & ICD-11
Items and clinical rating scales mapped to functional impairment criteria for Specific Learning Disorder and ADHD — not symptom counts alone.
04
IAP, IPS & DALI-DAB Indian Guidance
Aligned with Indian Academy of Pediatrics consensus, Indian Psychiatric Society guidelines, and DALI-DAB work on dyslexia assessment across Indian scripts.
05
Blinded Validation Study Design
100% of screen-positives, 50% of monitor-band, and 25% of screen-negatives receive independent blinded clinical assessment.

Ready to bring Kratu to your school?

Return to the platform overview or apply for school early detection below.