Drive — every file in the platform, one folder tree, no re-uploads.

The wedge

Every SaaS suite ends up with files scattered across ten places: attachments on tickets, PDFs on invoices, receipts on expenses, ID documents on employees. Odexy Drive is the folder tree that unifies them — every byte your other modules generated is reachable, shareable, and searchable in one place, without duplicating storage or breaking the source module's permissions.

What ships today

• Folder tree + explorer — vendored @reui/solution-files-1 canvas with folder rail + list/grid views + breadcrumb + drag-to-move.
• Files from everywhere — invoice PDFs, payslip PDFs, contract signed copies, expense receipts, ticket attachments, employee documents — every module's generated + user-uploaded files surface here without re-storage.
• Sharing — anonymous share links (HMAC-signed, expiring), per-user shares (editor / viewer / commenter), team shares, org-wide.
• Preview — inline preview for images, PDFs, CSVs, code, video, audio. WOPI-provider integration for Office / OnlyOffice / Collabora on .docx / .xlsx / .pptx.
• Co-editing — real-time TipTap for the platform's own document types; presence badges show every co-editor.
• Search — omnibox over filename + full-text where the extractor ran + tag matches + folder path.
• Revisions — file-level revision list per item; restore-from-here. Non-destructive.
• Trash + retention — soft-deletes retained per per-tenant slider; hard-delete after retention window unless legal hold.
• Comments + activity — per-item thread + per-item activity feed.
• Tag picker — per-org tag catalogue; tags drive both filtering and the "smart folder" view.
• AV + DLP gates — every uploaded byte hits an antivirus scan before it becomes reachable. scanState !== 'clean' refuses the storage.object.get_url call.

How it integrates

• Storage — Drive is a consumer of the unified Storage module. Same S3 / R2 / MinIO / B2 / Wasabi / DigitalOcean Spaces / Hetzner / Scaleway backends the rest of the platform uses.
• Sales / HRM / Contracts / Expenses / Support — the PDFs those modules generate + the files their users upload appear in Drive automatically. The source module owns write; Drive owns discovery.
• Chat + Mailbox — attachments dropped in chat + mailbox flow through Drive so they're recoverable without opening the original message.
• Recruitment — CVs uploaded during application land in the candidate's Drive folder; HR can browse them there without opening every candidate detail page.
• Automation — drive.item.created, .updated, .shared, .trashed register as automation triggers (auto-organize by kind, auto-notify on external-share, etc.).

How the AI helps

• "Find every invoice PDF Acme signed last quarter."
• "Summarize the last three revisions of the pricing deck I share with sales."
• "Move every screenshot in /inbox to /screenshots/2026-q3/."
• "Draft a share link for the design brief that expires in 7 days and needs a password."

The AI inherits your Drive permissions — if you can't see a file, neither can it. Share links it drafts go through the same drive.link.create action a human uses.

Built for scale

• BYOB (bring your own bucket) — pin the operator to their own AWS S3 / R2 / MinIO / B2 / Wasabi / DigitalOcean Spaces / Hetzner / Scaleway backend. Bytes never leave the operator's cloud.
• Storage profiles — per-org routing rules (this purpose → this bucket) for compliance-tier separation.
• Usage + quota — per-org total + per-purpose sub-caps. Nightly reconciliation cron catches drift between the meter + the provider's actual bytes.
• Object lock — compliance-tier surface for immutability + retention lock (SEC 17a-4, FINRA).
• Public / private / team — permission matrix per object; per-share audit trail.
• Multi-tenant safe — every path prefix keyed on org_id; no cross-tenant read is possible even with a leaked bearer token.