§ Radicept · AI-native DPDP compliance
It reads, maps, and drafts.
You keep the judgment.
An India-hosted AI maps your documents against frameworks that span India's data-protection laws and acts, finds the gaps, and drafts a regulator-ready report — with the evidence to prove it.
Dashboard
Monday, September 4, 2026128
Documents
96 indexed
12
Data Maps
34 activities
8
Frameworks
612 provisions
47
Findings
9 pending review
Compliance Score
Based on 47 assessed provisions across 8 frameworks
Status Breakdown
- Compliant 28
- Partial 11
- Non-Compliant 5
- Not Assessed 3
Recreated from the product interface
The problem
DPDP compliance still runs on spreadsheets, email, and PDFs.
Slow. Expensive. And impossible to prove after the fact.
Why existing tools don't solve it
Three dead ends.
Spreadsheets & consultants
Weeks of senior-lawyer time. It doesn't scale.
Generic GRC platforms
Checklist trackers that never read your documents.
General AI chatbots
Hallucinate the law. Send your data offshore.
The solution
It reads, maps, and drafts. You keep the judgment.
A five-stage pipeline turns raw documents into a defensible, regulator-ready posture.
- 01 Intake
- 02 Extract
- 03 Assess
- 04 Review
- 05 Report
How it works
A guided six-step workflow.
Client staff supply the inputs; the AI does the reading and mapping; the lawyer runs review and sign-off.
Questionnaire & intake
Client staff answer a questionnaire and upload documents themselves.
Document intelligence
Radicept reads every document — even scanned, image-only PDFs.
Data map
An inventory of data categories, activities and flows — built for you.
AI assessment & findings
Each activity is mapped to the applicable provisions and graded; gaps become findings.
Human review
The lawyer confirms, edits or overrides — only where it matters.
Report & evidence
A board-ready report and a dated evidence snapshot, in one click.
The foundation
The data map builds itself — but the risky calls stay rule-based.
The AI extracts entities from your documents to inventory categories, activities and flows. Sensitivity, though, is fixed by deterministic rule — so an Aadhaar or PAN is never a model's judgment call.
| Category | Sensitivity | Source |
|---|---|---|
| Aadhaar number | Special category | documents |
| Payment instrument | Sensitive | documents |
| Employee records | Sensitive | questionnaire |
| Marketing email | General | manual |
Sensitivity is set by deterministic rule — not model judgment.
What it brings to the table
The payoff.
Speed
Weeks of work compressed into hours.
Consistency
The same rigor every time — whoever runs it.
Defensibility
Cryptographic audit trail and dated evidence, on demand.
Scale
One lawyer serves many clients; staff do the legwork.
Why the approach is better
Not just AI — trustworthy AI.
Grounded, not generative
Cited provisions are checked against the statute corpus, and uncertain answers abstain rather than guess.
Hybrid, not black-box
Deterministic rules, retrieval and an LLM cross-check one another.
Self-calibrating
The confidence router learns from every lawyer decision.
India-native
Data residency and Indian-PII handling built into the core.
Measured, not claimed
Benchmarked, not asserted.
Every release is scored against a fixed golden-set benchmark before it ships — and the prompt that runs in the evaluation is the prompt that runs in production.
- Citations checked against the statute corpus
- Inference pinned to an India region in our deployment (§16)
- Aadhaar and PAN redacted from model prompts
- Prompt-injection and hallucination guardrails on model calls
Defensibility
Built to be provable.
Not just a verdict — a record. Each step is hash-chained, so the posture you report is the posture you can demonstrate.
- Tamper-evident audit log
- Hash-chained and append-only, enforced in the database.
- Evidence snapshots
- A dated, captured bundle of the evidence behind your posture.
- Human-in-the-loop trail
- The AI's verdict, recorded before the override.
- Integrity-checked exports
- Every export ships with a hashed manifest of its contents.
Dated bundles of this project's questionnaires, report summary and company documents.
- ViewPre-audit — Q3 2026 Report
Sep 4, 2026, 3:21 PM · by R. Mehta
3 questionnaires 24 documents - ViewBoard review Report
Aug 21, 2026, 11:04 AM · by R. Mehta
3 questionnaires 21 documents - ViewSnapshot
Aug 2, 2026, 6:47 PM · by A. Nair
2 questionnaires 18 documents
The moat
Why we win — and keep winning.
What a document-reading, statute-grounded, India-hosted system does that checklists and chatbots can't.
| Capability | Radicept | Generic GRC | AI chatbot |
|---|---|---|---|
| Reads your documents & Indian statute | Yes | No | partial |
| Citations checked against real law | Yes | No | No |
| India residency + PII scrubbing | Yes | rare | No |
| Lawyer-in-the-loop, overrides logged | Yes | manual | No |
| Cryptographic audit trail | Yes | basic | No |
Why now
The obligation is new. The capability to automate it is new. They arrived together.
An entire economy on a regulatory clock, ₹250 cr stakes, and India-first requirements global tools can't meet.
See Radicept on your own documents.
A guided demo on a real DPDP assessment — from intake to a regulator-ready report.