Corpus
Corpus is AI-assisted case-file and evidence analysis software built in-house by Dutch licensed investigation agency R.I.D. It breaks a complete dossier down into entities, a timeline, contradictions and evidence weighed per issue, with every finding traceable to its source passage.
What is Corpus?
Corpus is case-file and evidence analysis software developed in-house by the Recherche Inlichtingen Dienst (R.I.D.), a private investigation agency based in Amsterdam. Its name comes from corpus delicti, the body of the evidence, and that is what it assembles: it reads a complete dossier, whatever its format or size, and turns loose documents into one coherent, searchable picture.
The output is consistent across every field it addresses. Corpus identifies entities and the links between them — people, companies, accounts, telephone numbers and PGP identifiers, licence plates, locations — and builds a single timeline spanning all the material. It flags contradictions between statements and documents and sets them side by side. It weighs the evidence available for each element of a charge, each disputed point or each allegation, showing where a case is strong and where the weak link sits. Exculpatory material is surfaced deliberately rather than left buried. Every finding points back to the exact passage it came from, and a finding without a source document is not recorded at all.
Four legal specialisations are documented. Criminal defence gets an evidence matrix per element of the charge, signalling of procedural defects and support for formulating investigation requests. Civil litigation gets the file organised around pleading and the burden of proof, with evidence and contradictions per disputed point. Corporate disputes get ownership structures, money flows, contract clauses set against actual performance, and director liability. Employment matters get a dismissal or integrity file tested for completeness and consistency.
A separate configuration serves Dutch municipalities: a six-phase enforcement workflow with a legality check running alongside it, a bank-statement analysis module whose rules are configurable without code, signature comparison and document hashing, a report generator and a full audit trail.
Corpus is reached through the browser, signing in by QR code, passkey or a six-digit code from the RID Authenticator web app. The vendor states that personal data is pseudonymised before any model sees it, that customer files are not used to train models, and that the software produces no risk scores, no profiling and no automated decisions. The AI orders the material; the professional decides.
What it does
- Ingest a complete case file whatever its format or volume and make it searchable
- Extract entities and their links: people, companies, accounts, numbers, plates, locations
- Build one timeline covering every document in the file
- Flag contradictions between statements and documents and place them side by side
- Weigh the evidence available for each charge, disputed point or allegation
- Trace every finding back to the exact line of its source document
- Generate a report assembled from the steps actually recorded
When to use Corpus / When not to
A quick filter to help you decide if Corpus is the right fit.
When to use Corpus
- Criminal defence lawyers facing thousands of pages of police files, intercepts and seized chat datasets
- Civil, corporate and employment lawyers who must organise a file around the burden of proof before a hearing
- Private investigation agencies turning OSINT findings and surveillance reports into a structured, checkable dossier
- Dutch municipal enforcement and social-security investigation teams needing a legality check built into the workflow
- Corporate legal, compliance and internal audit teams running internal investigations, due diligence or fraud reviews
When not to use Corpus
- Organisations looking for risk scoring, profiling or predictive fraud detection, which the vendor refuses to build
- Anyone wanting to screen files or populations to find unknown suspects, rather than work inside an already opened case
- Teams that need automated decision-making: Corpus produces no decision and no decision proposal
- Practices outside the Netherlands, since the product and the whole site are built on Dutch law and are Dutch-only
- Users expecting self-service sign-up, published pricing or an API, none of which exist
How to use Corpus
A typical end-to-end flow, from setup to results.
- Contact the publisher by form or telephone: there is no online sign-up
- Attend a demonstration, after which a commercial proposal is drawn up
- Try a demonstration environment populated with fictitious data, at no commitment
- Run a pilot on a limited number of live cases within one team
- Have the action catalogue, calculation rules and report structure configured to your own procedures manual, in configuration rather than in code
- Install the RID Authenticator as a web app on your phone's home screen: no App Store or Play Store download
- Complete the first sign-in with a one-off pairing code scanned from the login screen
- Optionally register a passkey so Face ID, Touch ID or a fingerprint replaces the code
- Load the file from the exports, PDF attachments and overviews your existing systems already produce
- Work through the structured dossier, checking each finding against its source before it is retained
Pros & Cons
Pros
- Complete traceability: no finding is kept without a source document, and what the AI proposes is verified against the source first
- Profiling and risk scoring are refused by design, an argument the vendor grounds explicitly in the Dutch childcare-benefits scandal
- Personal data is pseudonymised before any model processes it, and customer files are not used to train models
- SHA-256 fingerprinting per document and a full exportable audit trail make integrity demonstrable months later
- Strong authentication without SMS: QR code, phishing-resistant passkeys and keys that stay on the user's own device
- Built by people who do the work: the publisher is itself a practising investigation agency, not a distant software house
- AI documentation is published unprompted, including a PDF written to be attached to a DPIA, and a demonstration then a pilot are offered before any commitment
Cons
- No published pricing at all: no pricing page, no amount, no tier, only a quote after a demonstration
- No API and no public technical documentation of any kind
- No named third-party integration; the publisher concedes that a real link to national systems does not yet exist
- Dutch only: the site, the product and the whole legal framing are Dutch, and no interface or processing language is declared
- The third-party AI provider is not named and no subprocessor list is published
- The hosting location is not published and is deferred to implementation, while transfers outside the EEA are accepted under standard contractual clauses
- No self-service trial and no permanent free plan, and the publisher is a very small operation with no social presence
Pricing & Plans
No price is published for Corpus. The publisher states that the amount depends on the number of users, the modules selected and the configuration work needed to match the customer's procedures manual, and that the structure combines a one-off set-up fee with an annual usage fee. A proposal is issued only after an introductory conversation and a demonstration, so no entry price can be quoted. No permanent free plan is announced, and the free demonstration environment and pilot offered beforehand are not described as a free trial. Readers should note that the amounts visible elsewhere on the site — a EUR 1,200 per year R.I.D. Membership and EUR 495 to EUR 695 voice-analysis tests — are agency service offerings and are not Corpus licence prices.
Data, GDPR & hosting
A consolidated view of how Corpus handles your data.
GDPR overview
The site carries one legal document, a "Beveiliging & privacy" statement for Corpus, version 1.0 dated June 2026; there are no terms and conditions and no separate legal-notice page. It claims processing under the GDPR, the Dutch Wpbr private-investigation act and the approved privacy code of conduct for private investigation bureaus. Concrete commitments are listed: a data processing agreement is concluded, DPIA documentation and a description of processing operations are supplied at implementation, transfers outside the EEA are covered by the European Commission's standard contractual clauses, data subjects may request access, rectification, erasure and objection through the published contact details, and personal data breaches are reported to the Dutch supervisory authority within 72 hours where required. No Article 27 representative is designated, the publisher being established in the Netherlands, and no data protection officer is named.
Who owns the data?
The vendor states plainly that the file remains the customer's: "the dossier stays yours", and, for investigation agencies, that the findings belong to their own bureau and not to a third party. Access is compartmentalised by case and by user, so an account only sees the files it has been authorised for, and account administration is reserved to an administrator. Temporary read access can be granted for demonstrations, but such a viewer only sees released demonstration files, never real client files, and can create nothing. An access log records who opened or edited which file, when and from which IP address; it is retained and can be exported for audit purposes.
Reuse rights
Corpus is presented as a working environment for a case that has already been opened, not as a data source the vendor exploits. Personal data — names, addresses, Dutch citizen service numbers and account numbers — is replaced by neutral labels before anything is processed, and the vendor states that the mapping between label and person stays in its own database while the result is reassembled inside Corpus. Text analysis is passed to a specialised third-party AI service, which is not named on the site, under business terms in which submitted data is not used to train models; the vendor adds that it supplies no more data than the analysis requires. Every document receives a SHA-256 fingerprint and every action is logged. The customer keeps and reuses its own findings without asking permission.
Data retention & training
Hosting summary
The publisher states that the data sits in its own environment, stored and transmitted encrypted, segregated per case and per user, with access always through multi-factor authentication. Beyond that, the hosting location is not disclosed: asked directly where the data is held, the site answers that this is recorded at implementation alongside the data processing agreement, the retention periods and the description of processing operations. No country and no region are named anywhere on the site. Text analysis is passed to an unnamed specialised third-party AI service, and the security statement accepts that transfers outside the European Economic Area may occur, covered by the European Commission's standard contractual clauses. The claim that nothing is kept in the cloud concerns the authenticator keys held on the user's own phone, not the hosting of case files. A prospective buyer should treat the jurisdiction question as open and settle it contractually.
Things to keep in mind
Risks and trade-offs to weigh before adopting Corpus.
- The POB 1290 ministerial licence, the police screening and the Wpbr framework accredit the R.I.D. agency, not the Corpus software: it is easy to read one as the other
- The EUR 1,200 per year R.I.D. Membership does not grant access to Corpus; it buys agency investigation services at a discount
- The third-party AI provider that performs the text analysis is never named and no subprocessor list is published, so the processing chain cannot be audited from the site
- Transfers outside the European Economic Area are accepted under standard contractual clauses, and the hosting location is negotiated at implementation rather than published
- The publisher argues that Corpus is not a high-risk AI system under the EU AI Act while explicitly leaving that assessment to the customer's lawyer: it is a vendor position, not an official classification
- The environment is working storage, not an archive: files are removed once a case closes, so customers must have their own archiving arrangements
- A tool that structures a dossier this efficiently can encourage professionals to accept its ordering without rereading the underlying documents, which is precisely the reflex the traceability design exists to prevent
Setup & Integrations
Technical difficulty
Low for the end user, moderate for the organisation. Nothing is installed: Corpus runs in the browser. First sign-in uses a one-off pairing code scanned with the RID Authenticator, added to the phone's home screen as a web app, with a passkey optionally registered afterwards; the publisher offers telephone help. The organisational work sits with the publisher, which configures the action catalogue, calculation rules and report structure to the customer's procedures manual, in configuration rather than code. Data arrives from existing exports and PDFs, as no automated connector exists. A pilot of a few weeks is recommended before wider rollout.
Deployment
Behind Corpus
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What exactly is Corpus?
Does the AI decide anything about people?
Are my case files used to train AI models?
How can I check that a finding is correct?
How do you sign in to Corpus?
Is there an app to download?
Is a data processing agreement available?
Where is the data hosted?
What does Corpus cost and can it be tried first?
Is there an API or an English version?
Should you pick Corpus?
Corpus is one of the more unusual entries in the legal AI field, because its publisher is not a software company at all but a licensed Dutch private investigation agency that built the tool for its own trade and then sold it on. That origin shows in the design. The refusals are as prominent as the features: no risk scores, no profiling, no automated decisions, no screening of files to find unknown suspects, and an explicit refusal to record any finding that cannot be pinned to a source document. The publisher grounds these choices in the Dutch childcare-benefits scandal rather than in marketing language, and backs them with pseudonymisation before processing, document hashing, a full audit trail and a written commitment that customer files are not used to train models. For lawyers and investigators who must defend how a conclusion was reached, that is the right set of priorities.
The reservations are commercial rather than ethical. Nothing about pricing is public: the amount depends on users, modules and configuration, and only a demonstration produces a proposal. There is no API, no named integration, no subprocessor list, no published hosting location, and the third-party AI service doing the text analysis is never identified. The product exists only in Dutch and is built entirely around Dutch procedural law, which caps its relevance sharply at the border. The publisher is also a very small operation with no social presence and a single legal document dated June 2026.
For a Dutch law firm, investigation bureau or municipality drowning in a large file, Corpus is a serious and unusually candid proposition. Outside that context it is hard to evaluate, and a buyer should expect to ask directly for the hosting, subprocessor and pricing answers the site does not give.
- Choosing a selection results in a full page refresh.
- Opens in a new window.