For teams
For 10–100-person companies whose knowledge lives in PDFs, spreadsheets, and email threads with suppliers, regulators, and customers. Tenant-isolated, EU-hosted, every answer cited.
Pilots run 4–8 weeks · 10–25 users · 1,000–5,000 documents · scoped to one tenant.
The problem
Mid-size companies sit on years of operational knowledge — contracts, invoices, quotations, scanned forms, spreadsheet ledgers, and email threads with suppliers, regulators, and customers. Full-text search misses synonyms, struggles on multilingual content, and cannot answer questions like "show me invoices from last quarter for supplier X over €5,000."
Email is worse — searchable only inside the mailbox itself, and only by users who already know what they're looking for.
Archivez makes that corpus answer questions.
Real user journeys
Procurement officer
Archivez identifies the supplier, applies the date filter, and retrieves the relevant clauses across multiple PDFs. The answer arrives with [1] [2] [3] page-level citations.
Two clicks open the original PDF at the cited page. The answer is grounded; the source is one click away.
Tenant administrator
Filters by sender and date, runs semantic retrieval against indexed email bodies and any PDF or Excel attachments.
If multiple people named Marcus exist, the system asks "which Marcus?" with a candidate list — instead of guessing.
Field engineer
Opens the mobile app at the customer site, asks the question in plain language, gets the answer plus a linked PDF in seconds.
Native iOS app, not a wrapped web view. Built for the read-quick / verify-fast use pattern of field work.
Operations lead
Upload the tender. Three minutes later, ask for every penalty clause — get a summary with each penalty page-cited.
Same flow for RFPs, regulatory submissions, statements of work, anything PDF-shaped.
Tenant isolation
Every row in every table is scoped by tenant ID. Every retrieval filters by tenant before the AI call. Cross-tenant data exposure is structurally prevented at the SQL layer — not by a checkbox a future ops mistake could clear.
Gmail OAuth tokens carry an additional KMS-wrapped envelope encryption layer. Document storage and AI inference are pinned to EU regions.
EU data sovereignty
eu-central-1 primary. Cross-region inference is pinned to EU regions only — Frankfurt, Stockholm, Milan, Spain, Ireland, Paris. No US fallthrough.
AWS Bedrock contractually does not train on customer inference inputs. We do not train any model on your data.
Data controller: Open Kinetix d.o.o., Serbia. DPA available on request. Standard subprocessor list.
Gmail OAuth tokens are wrapped with a tenant-specific KMS DEK on top of S3 at-rest encryption (AES-256).
gmail.readonly only. No send, modify, or delete. You pick which senders get indexed.
archivez.io ships no analytics, no third-party tracking, and no cookies. There is nothing for a banner to disclose.
See Security for the full technical detail, OAuth verification status, and CASA Tier 2 roadmap.
Business plan
€40 of AI usage included per month. Metered overage with a configurable cap. Annual prepay saves 15%.
Pilot program
Week 1
Tenant provisioned. Admins invited. Bulk-upload your sample document set.
Week 2
Documents finish indexing. Users connect Gmail. Monitored senders configured per user.
Weeks 3–6
Users ask questions, flag bad answers. Weekly check-ins on retrieval quality.
Weeks 7–8
Usage report (queries, citations, eval scores). Pricing for production rollout.
A 30-minute intro is enough to scope a pilot. Bring a sample of the kind of question you wish your documents could answer.