From demand letters to regulatory filings to arbitration submissions. Prilex reads your case files, extracts every fact, and builds the complete picture with citations to source, zero data retention by architecture.
Automated extraction, entity mapping, and chronology building for litigation, arbitration, investigations, and due diligence. One engine, every practice area.
From demand letters to trial prep: extract claims, map timelines, and build deposition-ready chronologies from any case file.
Learn moreProcess submissions, map multi-party relationships, and build the tribunal's working chronology, with every fact traced to its exhibit.
Learn moreRegulatory, internal, or law enforcement. Surface patterns across hundreds of documents, map entities, and identify contradictions automatically.
Learn moreAccelerate M&A and compliance reviews: extract obligations, flag risks, and build structured summaries from contracts and corporate records.
Learn moreNo implementation team. No 6-month rollout. Drop your files in. Start asking questions.
PDFs, scans, emails, contracts. Drag them into the desktop app. OCR and embeddings run locally on your machine. Nothing leaves until you query.
Events, obligations, claims, money flows, entities. Extracted in parallel, cross-validated, and citation-grounded to source text. Autonomous agents build the case map in the background.
Natural language queries, Deep Analyze for comprehensive review, cross-document timelines, and entity dossiers. No training required.
Click the tabs to explore. This is the real Prilex interface with a live case file — David Mattson v. Notareal Ltd.
| Legal Obligation | USD | Basis |
|---|---|---|
| Termination notice period salary | 194,040 | 12 months × USD 16,170/month |
| Severance payment | 194,040 | 12 months' salary as per contract |
| Statutory vacation pay | 94,000 | 123 accrued vacation days |
| Unpaid sales commissions | 6,000 | Work already performed |
Three architectural choices that make Prilex fundamentally different. Three capabilities those choices unlock. Each would require a legacy legal AI player to rebuild from the ground up. External inference routes through dedicated LPU hardware (Groq) with no writable storage: no KV cache, no persistent memory. Our inference partner operates pure compute infrastructure. They do not train models, do not store prompts, and architecturally cannot retain data. If the ZDR endpoint is unavailable, the request dies; it never falls back to a logging path. Documents stay on the lawyer's machine. OCR, parsing, and embeddings run locally. Your firm's orchestrator lives on a dedicated VPC in the jurisdiction you choose: Frankfurt, Helsinki, or your own colo. Never multi-tenant. Every query logged with model, prompt hash, tokens, latency, ZDR status, and source documents touched. Exportable as a signed PDF. The same artifact answers your CISO's procurement review and a regulator's discovery request. LPU inference delivers responses at interactive speed, not the 20-30 second waits of general-purpose cloud AI. Legal work demands focus, not spinners. Drafting, clause review, and document Q&A feel instantaneous. Each extracted fact carries a verbatim quote from the source. Click a claim, see the exact paragraph, page, section, document. Citations are generated by retrieval, not the model, so they're verifiable, not invented. Background agents map relationships across your matter, surface contradictions between documents, track obligations to their dates, and resolve entities canonically. The output isn't an answer; it's a case map. Drop your files and Prilex produces a full executive summary: case classification, parties with roles, claims ranked by evidentiary strength, critical dates with dispute flags, evidentiary gaps, and a recommended next step. Every finding traced to its source. Parallel extraction, embedding, and annotation pipeline processes 500,000 pages in 8 hours on cloud infrastructure. What a 20-person legal team takes months to review, Prilex handles while you sleep. Per-document cost: under 20 cents. Prilex doesn't just read each document; it connects them. Entity resolution across files, contradiction detection between accounts, and cross-document timelines mean you see the full picture, not isolated fragments. Here's what happens when you drop a demand letter, contract, or court filing into Prilex. One document, five structured views. Events, obligations, claims, money flows, and entities. All extracted in parallel, cross-validated, and citation-grounded to the source text. "ACI" in the demand letter, "Arctic Circle Innovations" in the contract, and "Arctic Circle Innovations Ltd" in the court filing. Prilex knows they're the same entity. One click to see every event involving them across your entire case. One click on any document for a comprehensive legal analysis: executive summary, chronological timeline with ALLEGED vs ESTABLISHED distinction, claims table, obligations, financial summary, strengths and weaknesses for both parties, and recommended actions. Every byte of your data stays in infrastructure you control. Inference runs on ZDR-by-architecture endpoints, not contractual promises. Every call is logged and auditable. Documents, OCR, and embeddings stay local. Nothing leaves your device until you query. The desktop app is your first and last line of defense. A dedicated orchestrator on infrastructure you control. Routes queries, enforces ZDR, and logs every external call. Per-customer, single-tenant, in your jurisdiction. Two-tier: LPU processors for real-time work (no KV cache, no disk, stateless by design) and Fireworks ZDR enterprise for heavy reasoning with DeepSeek V4. Both are ZDR by architecture, not by contract. Leading legal AI tools route through shared cloud infrastructure. Here's the honest comparison with our approach.Built for the work that matters.
ZDR by architecture, not by clause
Your hardware, your VPC, your jurisdiction
An audit trail you can hand a judge
Sub-second responses, always
Every claim, sourced to the line
Agents that build the case, not just answer it
Case Overview — the complete picture
500,000 pages overnight
Cross-document intelligence
What the desktop app actually does
Multi-view extraction
Entity resolution across documents
Apr 2018 · Payment obligations arise
Mar 2026 · Demand letter sentDeep Analyze
Admissible. Auditable. Defensible.
Your machine
Your private VPC
ZDR inference endpoints
Not another chatbot. A case intelligence platform.
Prilex Legal AI Platforms Data privacy model ✓ Architectural ZDR ✗ Contractual ZDR only Infrastructure Per-customer isolated VPC Shared multi-tenant cloud Model training on your data ✓ Impossible by architecture ✗ Opt-out only Platform Native desktop + web Web browser only Speed (interactive) Sub-second streaming ~50 tok/sec (cloud GPUs) Citation grounding ✓ Verbatim quotes with offsets Sentence-level footnotes Price $249/user/mo (annual) $1,000–1,200/seat/mo Minimum commitment None — cancel anytime $30K–$60K/year Onboarding ~5 minutes, self-serve 3–6 months, high-touch Contract Month-to-month Annual Audit trail ✓ Signed PDF export Internal logging only