Skip to content

Knowledge with a source.

A Polish graph engine and artificial intelligence, powering three products: LexArch for lawyers, Aesculapius for healthcare and ChainScan for analysing crypto-asset flows.

Book a presentation See the products
LexArch: AI for lawyers without disclosing client dataProductEvery conclusion comes with its sourcePrincipleThe graph engine: the shared core of every productTechnologyAesculapius: the patient's full therapy before you decideProductDecisions stay in human handsPrincipleChainScan: where the funds went on the blockchainProductDeployed on the client's infrastructureSecurityA full audit trail of every querySecurity
Products

Three products, one engine

Each product solves a specific problem in its field. All of them share the same core: a knowledge graph and an AI layer.

01 · LawLexArchPseudonymisation of legal documents. The lawyer works with AI, and client data never leaves the computer.Lawyers, law firms, public offices02 · HealthcareAESCULAPIUS.A clinical decision support system. Medication review on official data, with pharmacist oversight.Hospitals, clinics, IT vendors03 · Finance and AMLCHAINSCANA console for analysing crypto-asset flows across many networks, with risk scoring and a report for the file.AML obliged institutions
All products run on one graph engine. Data entered once connects with every later analysis.How the engine works
LexArchProduct 01 · Law

AI for lawyers, without disclosing client data.

A desktop application that detects personal data in legal documents locally and replaces it with consistent pseudonyms. Only text free of that data reaches the AI model, and the answer comes back citing the provision and the document.

For
Lawyers, law firms, legal departments, public offices
Platform
Windows · macOS · Linux
Data
Stored locally, AES-256 encryption
LocalDocuments do not leave the computer without pseudonymisation. Works offline and on-premise.
ConsistentThe same person is always [PERSON_1], across the whole matter and in every document.
VerifiableAI answers cite the provision and the document. The lawyer approves every result.
LexArch · chat in the context of a matterSample data
LexArch
MattersAnonymiseChatFilesSettings
Recent conversations
Claim maturityAppeal deadlineContractual penalty vs interest
Local · AES-256Data protection on
Claim maturityMatter: Action for payment · 12 documents
On the organisation's server
Is [PERSON_1]'s claim for repayment of the loan already due?

The loan agreement between [PERSON_1] and [ORGANISATION_1] sets no repayment date. In that case the claim becomes due after the agreement is terminated (Art. 723 of the Polish Civil Code), that is, six weeks after notice. [1]

The demand for payment dated [DATE_2] can count as notice if its delivery is shown. Acknowledging the debt in email correspondence interrupts the limitation period. [2][3]

ELI[1] Polish Civil Code, Art. 723DOC[2] Demand for payment.pdfDOC[3] Email correspondence.pdf
The AI model can make mistakes. The content must be verified by a lawyer.
Draft a statement of claimList the litigation risks
Ask a legal question…
Personal data will be replaced with pseudonyms before sending
Fig. 1. A lawyer asks about a matter, and personal data in the answer is replaced with pseudonyms.
AESCULAPIUS.Product 02 · Healthcare

Before adding a medicine, see the patient's full therapy.

A clinical decision support system. Medication review on official data, care after the visit and pharmacist oversight. The decision always stays with a person.

For
Hospitals, clinics, IT vendors (via API)
Used by
Doctors, clinical pharmacists, patients
Sources
Polish medicines register (RPL) · EMA · ATC classification
Identifies the medicineBy its entry in the official register, not by the name on the box.
Shows connectionsShared substances, ATC classes and contraindications, quoted from the SmPC.
Guides after the visitThe patient confirms doses, and their questions reach the doctor.
Aesculapius · checking a set of medicinesSample data
Spring BasilYOUR PATIENTPAT-CGLF-HHJG · 8 medicinesMedicinesConfirmations
result for 8 medicines · 28 pairs checked4 issues to review✕2 contraindications per SmPC!2 repeated ingredients
✕clarithromycin+ivabradinepending pharmacist confirmationin medicines: FromilidKlacidRaenom

“Concomitant use with ticagrelor, ivabradine or ranolazine is contraindicated.”Fromilid SmPC, section 4.3

!paracetamolin 4 medicinesAPAPAPAP 325 mgParacetamol MylanPanadol

The same ingredient 4 times: doses add up. For review by a doctor or pharmacist.

patient's medicines · 8
Fromilidclarithromycin✕ 2! 1
Klacidclarithromycin✕ 2! 1
Raenomivabradine✕ 1
Zocor 10simvastatin✕ 1
Panadolparacetamol! 1
+ 3 more medicines
What this check does not coverThe patient's condition, doses, and interactions outside the SmPC. Readings await pharmacist approval.
Fig. 2. The result of a medication check: contraindications quoted from the SmPC, and repeated ingredients.
CHAINSCANProduct 03 · Finance and AML

Where did the funds go? Across many networks at once.

A console for analysing crypto-asset flows. Paste a wallet address to see a transaction graph across networks, a reasoned risk score and a report ready for the file.

For
AML obliged institutions and the justice system
Deployment
On the institution's infrastructure (self-hosted)
Networks
7 in the demo version, 21 in the full version
See the flowTransactions from many networks in one graph, with amounts in USD and bridge crossings.
Assess the riskA score of 0–100 with a list of signals: OFAC sanctions, mixers, peeling chains, layering, bridges.
Close the caseA PDF report and an evidence package with a SHA-256 checksum.
ChainScan · flow graph and risk assessmentScreenshots
ChainScan screenshot: graph of fund flows for a wallet addressChainScan screenshot: risk assessment of an address with a list of signals
Fig. 3. The graph of fund flows (left) and the risk assessment with its list of signals (right).
Engine

One engine. Many uses.

Every product runs on the same core. Data entered once connects with every later analysis, so the system knows more over time.

  1. 01Sources

    Collecting

    Documents, scans, databases, public registers, the blockchain, medical publications.

  2. 02Graph

    Connecting

    People, companies, accounts, substances and events become nodes, and the relations between them become edges.

  3. 03AI

    Understanding

    Language models read text, recognise entities and point out hidden patterns and anomalies.

  4. 04Result

    Decision

    A clear report with a visualisation and a link to the source of every fact.

Your own data and your own questions?The products are ready-made applications, but the core is universal. Any dataset where connections matter can be modelled as a graph: contracts, supply chains, registers, correspondence.
Describe your case
Security

Built to work in institutions.

For organisations that cannot afford a mistake or a data leak.

Deployed at the client'sInstalled on the client's infrastructure, with no data sent outside.
Full audit trailEvery query and every conclusion is recorded and can be reproduced.
Access controlPermissions down to a single document and record.
Human oversightThe system supports specialists, and every decision is made by a person.

We will show the products on your data.

We prepare the presentation for your area: law, healthcare or the analysis of financial flows.

Email
kontakt@dataengine.pl
Phone
[+48 000 000 000]
Address
[Ulica, kod, miasto]

We process your data in line with our privacy policy.