Help
Frequently Asked Questions
Answers to the questions legal teams ask most about Olympus — confidentiality, sourcing, limits, access, and integrations.
Is Olympus trained on our documents or matters?
No. All AI processing runs through a BAA-covered, Zero-Data-Retention gateway, and your data is never used to train models — full stop.
How does Olympus keep our data confidential?
It's built for a HIPAA / legal-confidentiality context: encrypted in transit (TLS) and at rest (AES-256), with secure sign-in, optional MFA, an email-domain allowlist, per-user isolation, signed expiring download links, deletion on request, and an append-only audit trail.
How do we know we can trust the output?
The assistant is always grounded in your documents, and legal claims come with inline citations pointing back to the exact source passage — so verification is one click away. The Beings work with strict evidence discipline.
What files and limits apply?
Olympus reads PDF, DOCX, DOC, XLSX, and PPTX files up to 30 MB each. Standalone chats cap at 10 attachments per message — for larger sets, work inside a Project, which is built for volume.
How do we get access?
Access is gated. Sign-up requires an approved / allow-listed email; your administrator controls the allowlist.
Does Olympus replace a lawyer or give legal advice?
No. Olympus supports professional judgment; it doesn't replace it. It isn't a lawyer, and Holmes in particular is careful to surface what the evidence shows without straying into legal advice.
Can we bring our own model?
No. There's no 'bring your own model' key — model choice is managed by the platform.
Which integrations are available?
Everlaw (ediscovery, per-user encrypted key), public case-law research, mail and cloud storage (Gmail, Google Drive, OneDrive, Outlook, SharePoint, Dropbox via Composio), and voice dictation. Integrations are optional.
Still have questions? Talk to our team.