Services
Five tiers. One chain of custody.
The service covers the entire life cycle of the case file and is delivered in five sequential stages, each preparing the ground for the next. Tiers can be requested separately or together.
Tier 0 · MICAREXPUF
Precision ingestion
Turning the raw archive, physical or digital, into an ordered and verified digital stream is the foundation of the entire process.
Physical documents
- Collection against inventory, in SCANALYTIX.AI-specific containers, from your premises.
- Removing staples, flattening documents, opening correspondence.
- Sorting and chronological ordering: ascending or descending, at your choice.
Digital documents
- Transfer exclusively through secure channels: SFTP, secure API or a dedicated client portal, encrypted with TLS 1.2/1.3.
- No e-mail attachments and no public file-transfer platforms.
- Full inventory on receipt. Media archives (video, audio, graphics) are collected, processed and inventoried separately.
Unbroken chain of custody: uniquely serialised seals, GPS-monitored transport, handover reports at every step, restricted-access areas and CCTV. Full procedures: Annex 1.
Tier 1 · SARAH
Preparation
This is not archiving but digital curation. Here the risk of human error is removed and the ground is prepared for precision analysis.
Double inventory
On receipt and after preparation. Full traceability for every document.
Image quality score
A numeric test per page: blur, glare, type too small. Illegible documents are identified before processing.
Cleaning and upscaling
Removing background noise and increasing resolution, so every detail remains legible.
Scanning at 450–600 dpi
Scanning resolution, adapted to the condition of the material.
Renaming and standardisation
Uniform naming under the SCANALYTIX.AI standard: date, issuer, subject, recipient.
Separate archives
The original, unprocessed version and the processed version. Both are kept, for compliance.
Every step is recorded in the audit trail, time-stamped and with the operator identified by name.
Tier 2 · ZAR
Processing
Static documents become structured, secured entities ready for querying. The stage where the archive becomes a database.
- Conversion
- Transforming documents from the formats received into the formats optimised for analysis.
- BATES numbering
- A unique, sequential identifier per page, the international litigation standard: identifiable before the court.
- OCR
- Optical character recognition. Printed text from scans and PDFs becomes editable and searchable.
- OCR validation
- Identifying reading errors, using a dual engine.
- Pseudonymisation
- Direct identifiers are replaced with an artificial identifier; the process is reversible only through your organisation’s encryption keys. Recommended before any dialogue with an external language model.
- Anonymisation (removal of direct identifiers)
- Direct identifiers are permanently removed together with the related metadata. Irreversible. Note: removing direct identifiers does not automatically amount to anonymisation within the meaning of the GDPR if the person remains identifiable from context.
- Human in the loop
- An operator’s correction of data below the confidence threshold generates a new document, preserving the original and logging the changes.
- Source weighting
- Weighing the documents, required for the Tier 3 analysis: “Merge, Split & File Compression”.
- Case-specific metadata manual
- Key words and semantic expressions, agreed together with you, for the specific strategy of the case.
- Deterministic search and identification (dtSearch)
- Full-text indexing of the entire archive and reproducible search reports: every occurrence of a term, name, amount or phrase, with the document and BATES page where it appears. Exact, fuzzy, proximity or pattern search. The same query on the same index always produces the same result, independently of any AI model.
- Deterministic extraction
- Extraction of values (dates, amounts, parties, document numbers, issuers, etc.) based on predefined rules and validations, without a generative model. The result is reproducible and auditable; values below the confidence threshold go through Human-in-the-Loop correction.
- Granular encryption
- Differentiated access for departments, internal teams or external collaborators.
- Audio transcription
- Converting audio and video recordings in the file, provided by you, into text: provided you have the legal right to hold and process them. Verifying that right is your responsibility.
The result: a structured, numbered, indexed case file, ready for analysis.
Tier 3 · ZARIA
Analysis
From manual search to a strategic dialogue with the case file, in line with the EDRM (Electronic Discovery Reference Model).
Summaries
A document summary of variable length (essence, medium, detailed) centred on the topics you request.
AI-assisted extraction
From structured, semi-structured (forms, financial statements) and unstructured files, for the overall analysis of the case file. It complements the deterministic extraction in Tier 2; every value indicates the BATES page it comes from.
Identification
Establishing the relevant data sources: e-mail servers, workstations, cloud, mobile devices, at your request and within the limits of the legal basis available to you.
Legal hold
Ensuring data is not deleted or altered; any notices to the persons involved are issued by you, on the applicable legal basis.
Collection
Transfer in a secure environment, from sources to which you have a legal right of access, with metadata intact: creation date, author, change history.
Comparative analysis
Identifying contradictions between statements and the other documents in the file.
Review
Assessing documents for legal relevance and for attorney–client privilege.
Interactive dialogue with the case file
State-of-the-art models, accessed from your isolated environment; processing takes place at the provider named in the contract, on EU infrastructure, and only if you opt for this component.
Multimodal conversation
Topic-specific text reports, podcast-style audio synthesis, interactive Mind Map.
Dialogue with the case file
The basic rule of analysis: the system cites rather than fills in, and every answer can be checked at page level.
Anchored exclusively in the processed sources
- The system is configured to answer exclusively from the processed and uploaded documents, with a reference to the BATES page from which each statement originates.
- If the information is not found in the sources, the system is instructed to say so explicitly, rather than fill in the gap. The BATES references let you verify any answer quickly; verification remains necessary.
- Sources admitted simultaneously: the entire case file, with no contractual limit on the number of files or queries, plus external sources you define specifically.
Professional judgement and responsibility remain yours.
Conversation manual
Agreed together with you: questions are calibrated for the case strategy and to eliminate ambiguous phrasing.
Resumption brief
At every new hearing: a summary of the latest hearings and the vulnerable points identified.
Correlation with legislation and case law
Through the sources your organisation already subscribes to, where the provider allows technical integration. We do not resell licensed legal content.
Using an external language model (AI) is optional and your choice. Tiers 0–2 are delivered in full, without any call to an external provider.
EDRM
The nine SCANALYTIX.AI reference stages, EDRM-compliant
The Electronic Discovery Reference Model (EDRM) is a conceptual framework that standardises the steps for managing digital evidence (electronically stored information (ESI)) during legal proceedings and investigations.
- 1
Information governance
Managing the creation, storage and disposal of data before litigation begins.
- 2
Identification
Locating potential sources of relevant data and their custodians.
- 3
Preservation
Protecting data against deletion or alteration.
- 4
Collection
Gathering data while maintaining the chain of custody.
- 5
Processing
Converting files into a reviewable format.
- 6
Review
Evaluating documents for relevance, confidentiality and privilege.
- 7
Analysis
Evaluating content for key patterns, context and deeper insight.
- 8
Production
Delivering data to the opposing party.
- 9
Presentation
Presenting evidence at hearings, depositions or trials.
Tier 4 · REBEFRE
Storage
The data life cycle between hearings: low cost, maximum security, immediate reactivation.
Archive storage
Low costs between hearings, with government-grade security.
Encryption at rest
Content cannot be accessed without your organisation’s keys.
Instant reactivation
At a new hearing, the case file is immediately queryable again.
Audit trail
A complete log: who queried what, and when. Essential for compliance and for the profession’s ethical rules.
What this service is not
SCANALYTIX.AI is not an accredited electronic-archive administrator. The storage we offer is secure storage, not electronic archiving with legal value in the sense of the legislation on archiving documents in electronic form. If you need archiving with legal value, we will point you to an accredited provider.
The AUDIT TRAIL procedures are presented in Annex 2. Security
The archive assessment
A report describing what you have, what can be processed, how long it takes and with what risks: the document on which we commit to a firm deadline and volume.