CaseLens
On-Prem Legal Discovery on Local AI Models. Your Evidence Never Leaves Your Network.
CaseLens is an on-premises legal discovery appliance that runs entirely on local AI models — Ollama or LM Studio — inside your own network. No document, embedding, or prompt is ever sent to OpenAI, Anthropic, or any outside vendor. It ingests ESI from 15+ sources, processes 120+ file formats, and uses your local models to classify, cluster, and review documents at scale: hybrid keyword + semantic search, TAR/CAL predictive coding, and court-defensible production. Deploy it behind your firewall or fully air-gapped, sized to your matter volume — end-to-end discovery in one platform, with zero data egress.
System Capabilities
Local Models, Zero Data Egress
Every AI operation — classification, embeddings, semantic search, TAR/CAL, and RAG — runs on local models served by Ollama or LM Studio inside your network. No document text, prompt, or embedding ever leaves your infrastructure for OpenAI, Anthropic, or any cloud API. Run it fully air-gapped and the appliance still does everything.
On-Prem Appliance Deployment
Delivered as a self-contained appliance you run on your own hardware (or a Dynamanic-supplied box), sized to your matter volume and reviewer count. No tenant in someone else's cloud, no shared infrastructure, and no per-token or per-GB cloud bill — the model inference is yours.
ESI Ingestion from 15+ Sources
Pulls from Exchange, Microsoft 365, Google Workspace, Slack, Teams, SharePoint, OneDrive, Box, Dropbox, and email archives like Mimecast, Proofpoint, and Enterprise Vault into the on-prem store. Incremental sync with chain-of-custody logging built in.
120+ File Formats Supported
Processes email containers (PST, OST, MSG, NSF), Office and OpenDocument files, PDF and OCR images, audio with speaker diarization, forensic images (E01, AFF4, AD1), mobile backups, and chat exports from Slack, Teams, and WhatsApp.
Hybrid Keyword + Semantic Search
PostgreSQL FTS combined with vector search over locally-generated embeddings. Boolean operators, proximity queries, field filters, wildcards, and defensible search-term hit reports with CSV/XLSX export — all computed on your own hardware.
AI Reviewer Agent + TAR/CAL
A dedicated AI reviewer role with case-specific instructions for responsiveness and privilege, running on your local models. Continuous Active Learning predictive coding, RAG queries with source citations, and conflict detection across near-duplicates, families, and threads.
Court-Defensible Production
Bates stamping with custom prefix and numbering. Redaction with PDF burn-in. Metadata scrubbing for author, company, and revision history. Load file export to DAT, CSV, and Opticon. Defensibility packets with full audit trail.
Enterprise Controls and Compliance
AES-256-GCM encryption at rest. Tamper-evident audit hash chain. Legal hold automation with reminders and acknowledgment tracking. MFA and SSO. Because everything runs on-prem, matters are automatically compliant with GDPR, CCPA, and client no-egress mandates.
Deployment Process
Deploy the Appliance
We install CaseLens on your hardware (or a supplied appliance) inside your network, point it at your local Ollama or LM Studio model server, and connect your data sources. Air-gapped deployments are fully supported — no outbound internet required to run.
AI Processes and Reviews — Locally
The platform extracts text from 120+ formats, deduplicates by hash, identifies near-duplicates and email threads, classifies privilege and PII, and runs TAR/CAL on responsiveness — every model call served by your local models, never a cloud API.
Review and Produce
Reviewers code documents through the QC workflow. Bulk-coding propagates decisions to similar documents with rollback. Production exports load files with Bates stamps, redactions burned in, and metadata scrubbed.
Private Deployment
Custom pricing
On-prem appliance · annual license, sized to your matter volume · runs entirely on your local models, so there are no per-token or per-GB cloud charges.
Connects to Your Existing Stack
CaseLens plugs into the platforms you already run.
Common Questions
Does any of our data leave our network?
No. CaseLens runs every AI operation on local models served by Ollama or LM Studio inside your environment. Document text, embeddings, prompts, and model outputs never leave your network — nothing is sent to OpenAI, Anthropic, or any outside API. The appliance runs fully air-gapped if you require it.
Which local models does it use?
Any model you can serve through Ollama or LM Studio — Llama, Qwen, Mistral, and similar open-weight models for classification, embeddings, and review. We help you select models sized to your hardware and matter volume during deployment, and you can swap them later.
How is the platform court-defensible?
Every action is logged in a tamper-evident audit hash chain with HMAC signatures. Search hit reports, coding decisions, and production exports are reproducible. Defensibility packets bundle the full evidentiary record for court submission. AES-256 encryption at rest meets FIPS 140-2 standards.
What file formats can it process?
120+ formats including email (PST, OST, MSG, EML, NSF), Office and OpenDocument files, PDF, images with OCR, audio with speaker diarization, forensic images (E01, AFF4, AD1), mobile backups (iOS, Android, WhatsApp CRYPT), chat exports (Slack, Teams, WhatsApp, Zoom), and financial chat (Bloomberg, Reuters Eikon, Symphony).
Does AI review replace human reviewers?
No. The AI Reviewer Agent operates under a dedicated role with case-specific instructions and produces auditable markup — highlights, annotations, redactions. A human always validates production decisions. TAR/CAL accelerates first-pass review, then QC reviewers and senior reviewers handle escalations and final coding.
How does data residency work?
Your data lives wherever you deploy the appliance — your data center, your private cloud, or a fully air-gapped network. Because all processing and AI inference happen on-prem, every matter is automatically compliant with the strictest residency and sovereignty requirements (GDPR, CCPA, client no-egress mandates).
How is CaseLens priced?
CaseLens is an on-prem appliance, licensed annually and sized to your matter volume and reviewer count — not a per-seat cloud subscription. Because inference runs on your own local models, there are no per-token or per-GB cloud charges. Contact sales for a deployment scoped to your environment.
See CaseLens Running Live
We will show you CaseLens processing real data in a live demo. No slides. No pitch deck. Just the system running.
See It Run