Data security · AI · RDA · GDPR

Discover, understand
and control your
sensitive data.

Personal data lives everywhere — local drives, databases, cloud storage, legacy systems and employee machines. Most organizations simply don't know where it is. You can't protect what you can't see. LiveComply scans your entire environment, locates sensitive data, recognizes its type and context, links it to specific people — and hands you full control.

18+
Types of sensitive data detected
and classified automatically
0
Documents leaving your network
(on-premise deployment)
PL+EN
Full language coverage today
DE + FR on the roadmap
00 / Hero
RISK

You don't know
where your sensitive
data is — and that's
your risk

Every unknown copy of personal data is a potential breach, fine and reputational risk. Traditional DLP tools miss context — a name alone is not always personal data. And data grows every single day.

"Most organizations simply don't know where their personal data is. Security starts with answering that question."

Data sprawl
Employees store sensitive documents on local drives, in email and on shared resources.
Identification
Personal data is embedded in unstructured text: contracts, judgments, HR documents.
Scale
Thousands of documents per organization — and more every day.
Accountability
Security and GDPR both require knowing exactly where every piece of data lives.
01 / The Problem
CONTROL

One platform,
three capabilities.

01 — DISCOVER
Locate it

Harvester agents scan every data source — files, databases, cloud, employee drives. A full map emerges: where data is in the first place.

02 — UNDERSTAND
Know what & whose

Multi-layer AI identifies, classifies and links data to people. You know not just that you hold data, but what kind (Article 9 special categories) and whose.

03 — CONTROL
Act on it

Automated compliance workflows and data subject rights. You decide: encrypt, anonymize, flag, delete. Data becomes a controlled asset.

✓ On-premise or SaaS — documents needn't leave your infrastructure ✓ Works with existing document infrastructure — no migration ✓ Polish & English today; German & French on the roadmap
02 / The Solution
PROCESS

From data source to decision,
fully automatic

Harvester Queue NER gate Expert rules LLM Dashboard
  • The NER pre-scan is the efficiency gate — documents with no sensitive data are skipped before any expensive processing runs.
  • A message bus decouples discovery from analysis — horizontal scaling.
  • Results are structured and published to the dashboard in real time.
  • The Harvester runs like an antivirus agent on employee machines, or as a cloud connector. No document is stored centrally.
// Where we look — Harvesters
📁 File system
PDF · DOCX · XLSX
TXT · CSV · images
🗄️ Databases
SQL Server · PostgreSQL
MySQL · Oracle · SQLite
☁️ Cloud
AWS S3 · Azure Blob
Google Cloud Storage
📂 Google Drive
Team drives &
shared folders
03 / How it works
ENGINE

Three layers of AI.
Speed, precision, understanding

LAYER 01
NER gate

A fast classifier that decides whether a document contains genuine sensitive data at all. Fine-tuned on a proprietary legal corpus, runs on CPU too, detects 18+ entity types with confidence scoring — eliminating the processing cost of most collections.

LAYER 02
Expert rule system

High-precision, fully deterministic extraction of structured identifiers. 18 rules combining patterns with checksum validation: national ID, tax/registry numbers, IBAN (Luhn), passport, driving license. No ML — faster, deterministic and fully auditable.

LAYER 03
Person grouping (LLM)

The hardest problem: resolving whom each piece of data belongs to. A dedicated Polish-first LLM links scattered fragments to individuals. Output constrained to a strict JSON schema — the model can't invent values. Temperature = 0 → deterministic and auditable.

// THE PROPRIETARY DATASET — A MOAT

~2,000 documents (5+ pages avg): real, court-anonymized judgments plus synthetic agreements with realistic but fake data (valid national-ID checksums, correct address formats). Multiple rounds of expert annotation with relationships annotated by hand — which is why the system links scattered data back to a person.

// WHAT YOU GET BACK

For each document — machine-readable persons with fully linked data, split into GDPR categories (PersonalData and Article 9 SpecialCategoryData: health, biometrics, ethnic origin, views). Every entity carries its exact position, context, confidence score and identifier.

04 / Technology
ACT

Three ways to
take control.

Continuous monitoring

Scheduled scans across teams, drives and databases. Track data-exposure trends over time. Automated alerts on new findings and policy violations.

Ad hoc scanning

On-demand sensitive-data reports for any document set. A full audit trail: what was scanned, when, and what was found.

Data subject rights

GDPR Art. 15/16/17 in practice:

  • Access — find every record about a person across all systems
  • Rectification — locate and flag incorrect data
  • Erasure — identify every copy for deletion

"Data security stops being a manual, legal burden — it becomes an automated, auditable process."

05 / Dashboard
SECURE

A tool that maps your data
must not become a risk itself

🏢 On-premise
Harvesters on local network → internal message bus → PII detector + LLM (local GPU) → API + dashboard.
Documents never leave the network. Zero data egress.
☁️ SaaS
Harvesters locally → encrypted transport → managed LiveComply cloud processing → hosted dashboard.
No local infrastructure required.
🔀 Hybrid
Harvesters on-premise, processing in the cloud.
The best of both worlds.
06 / Deployment
PL · EN

We start with the
hardest language

The architecture is language-agnostic: swap the LLM backend per language, extend expert rules per jurisdiction, retrain NER on a new corpus.

"Polish legal NLP is among the hardest challenges in European compliance. We solved it first."

🇵🇱 Polish ✅ Full coverage Dedicated LLM, full legal entity taxonomy
🇬🇧 English ✅ Full coverage Included in training corpus
🇩🇪 German 🔜 Planned Largest EU economy, strong compliance culture
🇫🇷 French 🔜 Planned EU institutional and enterprise market
07 / Languages
FOR WHOM

Who is the
product for?

Data Administrators
& Data Protection Officers

Locate and secure every set of sensitive data — across internal environments, cloud and multi-cloud. Complete visibility, zero blind spots, one source of truth.

DATA PROTECTION · AUDIT READY · DPO TOOLS
Enterprises
& Financial Institutions

Know your data and identify relations between individuals across environments — private clouds and corporate servers. Full KYC & AML compliance at scale.

ENTERPRISE SCALE · KYC · RISK MANAGEMENT
08 / Audience
NOW

A proven foundation, a new
layer, an accelerating market

The foundation is proven

Years of real annotation, multiple model iterations, tested on real legal corpora, production-deployed.

The LLM layer is new

A dedicated LLM replaces the brittle relation model; structured output removes hallucination risk.

The market is accelerating

€4.2B in GDPR fines since 2018; data sprawl grows with cloud adoption; no dominant automated solution for European compliance.

€4.2B
GDPR fines issued
since 2018
4%
Max fine of global
annual turnover
€20M
Or maximum penalty,
whichever is higher
09 / Why now
VOICES

What users say
about LiveComply.

"Now we can comprehensively examine the legal situation regarding GDPR by devoting only a fraction of the time of a traditional audit. This saves costs for us and the client — a significant competitive advantage."

ŁK
Ł. Kulicki
CEO · Law & Consulting Firm

"Extremely innovative and indispensable tool — automating the manual, time-consuming elements of personal data management. A genuine breakthrough for compliance teams."

ŁC
Ł. Czernikow
GDPR Initiatives Lead
10 / Testimonials
TEAM

The specialists
behind LiveComply.

GK
Grzegorz Kanka
CEO
KC
Krzysztof Cieśliński
CTO · Cloud Architect
AS
Dr. A. Smywiński-Pohl
AI · Research Engineer
PK
Dr. Paweł Kapłański
AI · Research Engineer
KW
Krzysztof Wróbel
NLP · ML Engineer

Top AI engineers · Extensive experience in AI-based solutions · Lawyers & Auditors

11 / Team
FAQ

Frequently
Asked
Questions.

Everything you need to know about how LiveComply finds, classifies and controls your sensitive data.

How do I know where my company keeps sensitive data?
+
That's what LiveComply is for. Harvester agents search drives, databases, cloud and employee machines, and a three-layer AI engine recognizes, classifies and links what it finds to people. You get a single map — where and what sensitive data you actually hold.
Does my data ever leave my organization?
+
It doesn't have to. In the on-premise model, documents are processed entirely within your network — zero egress. SaaS (encrypted transport) and hybrid models are also available.
Can the AI "invent" data that isn't there?
+
No. The LLM returns output strictly in a JSON schema and only links spans actually detected in the document. At temperature = 0, results are deterministic and auditable.
What data is detected?
+
Over 18 types: personal data (name, address, contact, national ID, tax/registry numbers, ID-card number) and GDPR Article 9 special categories (health, biometrics, ethnic origin, views) — along with the relations between data and individuals.
Does my company need to be GDPR compliant?
+
GDPR covers almost every organization in the EU, plus non-EU entities handling EU citizens' data. The maximum fine is 4% of global annual turnover or €20M, whichever is higher.
12 / FAQ

Take control
of your data
with Live.Comply.

Try the free trial version. No payment required.

GDPR Compliant
ISO 27001
AI-Powered
// New message