Data Lens finds the data spread across your drives, clouds, and applications, classifies what it holds and what it is for, then acts on it. You get back not just a map, but data that is ready for AI and under your control before a model ever touches it.
Start with a free assessment on a single source, and see it on your own data before you commit to anything.
Files pile up across shared drives, cloud stores and business applications faster than anyone can track. Nobody can say with confidence what is in there: which files hold personal data, which duplicate each other, which sit open to people who should never see them, and which quietly carry instructions written to steer a model.
Point AI at that, and every one of those unknowns becomes a live risk. Getting ready for AI starts with seeing clearly what you already hold.
Data Lens reaches across the places your data actually sits, from Google Drive and OneDrive to cloud storage, on-premise systems and business applications, and builds a single map of what you hold. Most teams meet the true size of their footprint for the first time here. It is usually far larger than the estimate they walked in with.
Over a hundred classifiers cover sensitivity, PII, sovereignty, duplicates, and your own custom patterns, at over 99% accuracy. Every detection carries a confidence score, so your team reviews what genuinely needs a second look instead of wading through alerts.
Most tools are built for the common, global identifiers and thin out on two kinds in particular, the ones specific to a region, and the noisy types where false positives are hard to control. Data Lens goes deep on both.
Take India. Data Lens reads sensitive types from PAN and Aadhaar to UPI IDs, bank account numbers, mobile numbers, ABHA health IDs, and TAN, where common tools stop at the obvious few. The same depth extends to every market we enter.
And it finds them wherever they hide. A GST number buried in a text file, an Aadhaar sitting in a spreadsheet cell, the places other tools scan straight past.
Coverage spans hundreds of types, including the local ones others skip:
A file can be spotless on content and still be built to make a model misbehave. Data Lens reads the intent inside your data: hidden characters, embedded instructions, and jailbreak payloads planted to hijack a model downstream. It tells you what is genuinely safe to put into a RAG index or a fine-tuning run, and what to hold back.
Zero-width and control characters slipped in to carry instructions a human reviewer never sees.
Prompt-injection text seeded in documents so a model reading them quietly changes behaviour.
Content shaped to break guardrails the moment it lands in a RAG index or a fine-tuning set.
Flagged as unsafe for Copilot, RAG, or agent ingestion. These files carry jail-break and prompt-injection patterns designed to hijack a model that reads them downstream.
Poisoned data is the one risk that gets worse the moment you use it. Data Lens catches it before it reaches the model.
Every classification becomes a label written into the file, not filed away in a report no one reads. The label travels with the data. At runtime, Cyvia Intent Gateway reads it back to decide what any prompt or agent is allowed to do with that data, and Intent Router reads it to keep sensitive work on an approved model. Classify once here, and that understanding carries into every decision downstream.
Understand your data once, at rest. Control it everywhere after.
Discovery has a habit of surfacing files open to the whole company, sensitive data sitting where it should not, and copies multiplying quietly. Data Lens lets you act on what it finds, right from the picture it builds: delete files, shut down sharing, and manage access down to the identity, linked to your directory. And because exposure rarely comes one file at a time, you can act on a whole set of detections at once rather than clearing them one by one.
Remove files that should not exist, straight from the finding.
Revoke anyone-with-link and over-broad access in place.
Control by person and group, linked to your identity provider.
Clear a whole set of detections at once, not one by one.
Access actions link to your identity provider, including Entra, so you manage exposure by person and group, not just by file.
Every DPDP obligation rests on three questions: where is your personal data, who can reach it, and can you prove it. Those are the exact questions Data Lens answers, by discovering and classifying personal and sensitive data across everything you hold.
Data Lens gives you the data foundation the regulation rests on: a live map of personal and sensitive data, the exposure around it, and the artifacts to evidence it, including RoPA and eight other DPDP reports generated from the same scan. It is the hardest part of readiness, handled.
We have the data covered. You take care of the rest. Data Lens handles discovery, classification and the reporting artifacts underneath your practice, while you own consent, programme management and advisory, and deliver a complete offering to your clients with the hardest layer already solved.
Cyvia covers the data side of DPDP. Consent management and full-programme compliance are delivered with our partners.
Same engine, same classifiers. You choose how much of the stack lives in your environment.
Fully managed on Cyvia cloud, live in days.
Single-tenant, inside your own cloud account and keys.
Runs entirely in your data centre. Data never leaves.
Fully isolated with zero outbound calls, for sovereign estates.
Connects to Google Drive, OneDrive, AWS and Oracle VM out of the box, with more on request.
The first scan runs overnight. By the next morning your data is discovered, classified, and ready to act on. From there, a short, structured rollout takes it to production.
Stand up Data Lens and run the first scan across a live source. The footprint appears.
Tune classifiers against your own data with human review, and shape the setup to your policies. Accuracy moves from high to precise.
Set your policies, and Data Lens writes labels and metadata into the files automatically. Turn on remediation.
Extend across sources and repositories, and hand the labels forward to runtime.
Point us at one repository, or a slice of it. We run the scan and show you what is there, discovered, classified, and ready to act on. The first look is free, and it tends to be the moment it becomes real.