01
Controlled access
Case Managers authenticate before using the dashboard. Participants use time-limited invitation links instead of creating accounts.
Trust Center
Sensitive information deserves clear boundaries.
Arvanic is built for workplace complaint records that employers and participants may treat as confidential. This page explains what the product does, where automation stops, and which service providers help operate it.
Product controls
Designed to support a careful employer process.
02
Tenant isolation
Investigation records are scoped to the authorized Case Manager and organization, with database-level safeguards designed to prevent cross-account access.
03
Source confidentiality
No participant learns who reported the concern or what any other participant said. Each account is given independently, which protects the integrity of the record and reduces the risk of retaliation.
04
Pseudonymization and data minimization
Before AI processing, Arvanic replaces known participant and Case Manager names with role-based placeholders and removes common email, phone-number, and Social Security-number patterns. This reduces identifying information but does not make the text anonymous; unmapped names and indirect identifiers may remain.
05
Not used to train AI models
The AI providers that process investigation text for Arvanic are contractually prohibited from using it to train their models. Your investigations inform your report and nothing else.
06
Human decision-making
Arvanic is designed not to score individuals, make employment decisions, or recommend employment actions. Human decision-makers remain responsible for evaluating the information and determining what action, if any, to take.
07
Neutral language
Reports set out what participants reported, not what it means. Conclusions, characterizations, and legal terms are kept out, so the document stays factual and appropriate for whoever reads it next.
08
Time-stamped activity records
Relevant platform activity is recorded with timestamps so an authorized reviewer can understand how the investigation record developed.
09
Immutable record
Investigation activity is written to a history no one can rewrite — not a Case Manager, not an administrator, not Arvanic. Nothing is edited after the fact to fit a later account, which is what lets the record hold up under scrutiny.
10
Restricted platform access
Investigation content is not open to Arvanic personnel for routine browsing. Access is deliberate and recorded together with the reason for it, so there is always an answer to what was viewed and why.
11
Purpose-bound retention
Investigation data is retained for 36 months after final closure, unless a hold applies. Case Managers can request earlier deletion of a closed investigation through a governed process with a seven-day reconsideration period.
12
Adult participants
Arvanic is built for workplace investigations involving adult employees. Case Managers confirm each participant is 18 or older, and online participants confirm for themselves. Arvanic does not verify age, so an investigation involving someone under 18 is handled outside the platform — a limit on the tool, not on any worker's rights.
AI boundaries
The software organizes information. People decide what it means.
Arvanic may summarize and organize collected information, identify differences or omissions, surface open factual questions, assist with bounded fact development, and provide policy-grounded procedural support.
Arvanic does not decide credibility, what happened, fault, liability, legal sufficiency, discipline, termination, or whether counsel is required.
Arvanic is not a substitute for legal advice or a law-firm investigation.
Safety screening
An automated flag is a prompt for human review—not a finding.
Arvanic screens written responses for language that may need human review. A severe automated flag can stop an interview and notify the assigned Case Manager; other flags may appear in Safety Review.
Automated screening can over-flag or miss language. A flag is not a factual, clinical, or legal conclusion. Arvanic is not an emergency, crisis, continuous-monitoring, or response service. In an emergency, contact local emergency services.
Service providers
The services involved in operating Arvanic.
These providers support specific parts of the platform. More detail about the information involved appears in the Privacy Policy.
AuthenticationClerk
Database and file storageSupabase
Application hostingVercel
AI text processingAnthropic and AWS Bedrock
One-time paymentsStripe
Transactional emailResend
Customer support emailGoogle Workspace (Gmail)
Error monitoringSentry
Rate limitingUpstash
Questions about security, privacy, or data handling?
We welcome specific questions. Please do not include participant information or investigation content in your first message.