Trust Center
Real status, with the evidence behind it: assurance status, the security package, processor terms and subprocessors in one place.
Where each framework stands, stated plainly.
- SOC 2In progress
- ISO/IEC 27001In progress
- CSA STARIn progress
- GDPRDPA available · subprocessors listed
None of the three is complete yet, and we do not show a badge for one we have not earned. While they are in progress we can share our current controls, policies and evidence under NDA. Request the security package. Roadmap Completing all three is on the Roadmap.
Every assurance claim, with its artifact, issuer, date and scope.
Each row names the document behind a claim, who issued it, when, what it covers and how to get it. A cell that is not filled in is not yet confirmed. Until then, rely on the status wording in the next section, not on this table.
| Claim | Artifact | Issuer | Date or period | Scope | How to obtain |
|---|---|---|---|---|---|
| SOC 2 | Report with the auditor’s opinion | In progress | Not yet issued | Not yet issued | Security package under NDA: request it |
| ISO/IEC 27001 | Certificate | In progress | Not yet issued | Not yet issued | Controls are mapped to ISO/IEC 27001:2022 Annex A |
| CSA STAR | Registry listing | In progress | Not yet listed | Not yet listed | Not yet listed |
| GDPR and DPA | Signable Data Processing Addendum with Standard Contractual Clauses | Araghatta LLC | Last updated 4 October 2026 | Platform data processed by Araghatta as processor | Read the DPA, or request a signed copy |
| Internal web assessment (DAST) | Summary of an unauthenticated assessment of the public site, with fixes and re-test | Araghatta (internal) | August 2026 | Public web estate and dashboard login surface, from the internet; the sensor binary is not covered | Request the summary under NDA |
| Independent penetration test | Test summary, once performed | Being commissioned | Not yet performed | To include the sensor and the collector | Not yet available |
| Sensor security whitepaper | Control-by-control description of the sensor | Araghatta LLC | Revised 4 October 2026 | Sensor data handling, transport, storage, permissions | Request with the pack |
| Subprocessor list | Dated list of third parties that process data | Araghatta LLC | 4 October 2026 | Managed deployments (in-VPC has none for data processing) | Open the list |
| Accessibility review | Accessibility statement (WCAG 2.2 AA target) | Araghatta LLC (self-declared; no external audit is stated) | Reviewed 4 October 2026 | Public website and product experience | Read the statement |
Framework status, stated plainly.
We publish our real status and the evidence behind it. Where a framework is a mapping or an alignment rather than a certification, we say so.
| Framework | Status | Notes |
|---|---|---|
| SOC 2 | In progress | Not yet complete. Current controls, policies and evidence are available under NDA. |
| ISO/IEC 27001 | In progress | Not yet certified. Controls are mapped to ISO/IEC 27001:2022 Annex A. |
| CSA STAR | In progress | Not yet listed in the CSA registry. |
| ISO/IEC 42001 (AI management) | Aligned | Our own product maps findings to 42001; internal alignment underway. |
| GDPR | Aligned | Signable DPA, published subprocessors; data-subject request process in place. See Privacy. |
| EU AI Act | Mapped | Findings map to Art. 10, 14 and 15, and Art. 4 literacy in the risk register; other articles can be referenced manually. Roadmap Automatic mapping for more articles is planned. As of 6 October 2026; mapping and orientation, not legal advice. |
An unauthenticated web assessment (DAST) of the public site was run internally in August 2026 (summary under NDA). An independent third-party penetration test, including the sensor and the collector, is being commissioned; it has not been performed yet. Roadmap See how we handle the privacy of monitored employees.
Everything a vendor-risk review asks for.
What the sensor collects, and what it never does.
What it collects
Metadata of AI usage: process/binary names, MCP config locations, API-key presence and provider (never the value), installed AI CLIs/SDKs, browser-extension IDs, and egress destinations to known AI endpoints, attributed to machine and user.
What it never collects
Not the contents of your files, not secret values (masked at the point of collection), not keystrokes, not stored prompt/response bodies, not personal browsing. Command lines are scrubbed of secret-shaped tokens before they leave the endpoint. The optional in-browser DLP extension checks submitted text in memory and does not store it.
Read-only discovery, opt-in enforcement.
Discovery is read-only
By default the sensor observes and reports. It does not modify files, install software, or change security settings.
Enforcement is available, opt-in and audited
Enforcement is graduated and criticality-gated, from warn to block or quarantine. It is built into every sensor, switches on only with the Enforcement licence and your opt-in, and every action is operator-authorized and logged. The sensor checks on the machine that the block or isolation is really in force and reports it. Server-side content-DLP decisioning is available through an inspection API. The optional in-browser DLP extension runs on user workstations, not servers. It sends the text a person is about to submit to ChatGPT, Claude or Gemini, over TLS, to your own Sargha collector for a block, redact or allow decision. The text is checked in memory and is not stored: only the decision, the destination site and the kinds of data found are recorded, never the text. It never goes to the reasoning model, and the sensor itself never reads prompts.
Deployment options
Managed (our AWS: each customer is logically isolated by default; a physically dedicated tenant is a custom project) or fully in-VPC: the console and the classification model run inside your AWS account with no data egress. In-VPC uses Amazon Bedrock for reasoning so prompts never leave your boundary. Roadmap A documented air-gapped, zero-egress package is planned.
Data residency
Managed tenants are hosted in the United States (AWS us-east-1). In-VPC keeps 100% of data in your account, in the region you choose. Roadmap Regional data planes, EU-region managed tenants and residency pinning are planned.
How access and secrets are protected.
In transit
Encrypted everywhere. Sensor-to-console traffic is encrypted, and each sensor is authenticated to your tenant.
At rest
Encrypted storage; secrets held in managed key-protected storage or your own key store.
Access control
Role-based access across four roles (Viewer / Analyst / Admin / Owner, plus a Platform Admin role for Araghatta staff) and custom roles, TOTP multi-factor authentication, and an audit log of security-relevant actions. SSO (SAML and OIDC) and SCIM 2.0 are live (Discover + Warn + Block). Sign-in can be limited to the email domains you list. Temporary Admin access needs a reason and a different Owner's approval, and ends by itself. Roadmap Customer-managed keys on managed tenants are planned.
Integration secrets
Integration secrets (Slack webhooks, Jira tokens) you enter are encrypted at rest and never returned to the browser.
A short, current list.
We keep this list short and current. Managed deployments use:
| Entity | Purpose | Data | Location |
|---|---|---|---|
| Amazon Web Services | Hosting, storage and secrets (KMS) for managed deployments | Tenant console data and the masked findings the sensors report (encrypted at rest); no file contents, secret values or stored prompt/response bodies | Hosted in the United States. United States only: AWS us-east-1 (N. Virginia). Managed tenants are not offered in an EU region today. Customer-cloud (in-VPC) deployments run in the region you choose |
| Google Workspace (SMTP relay) | Transactional email (enquiries, alerts) | Recipient addresses and the content of enquiry and alert emails | United States |
| Reasoning backend (configurable): OpenAI (default) | AI triage and explanation of findings: a managed hosted model (OpenAI, United States by default), or Amazon Bedrock inside your account (in-VPC), or fully local inference; deterministic-only sends nothing to a reasoning model | Metadata about a finding, such as its title and the host it concerns. No file contents, secret values or prompts. Not used for model training; the provider may retain inputs up to about 30 days for abuse monitoring | United States by default; inside your account for in-VPC; none for local or deterministic-only. Provider region: United States |
In-VPC deployments have zero external subprocessors for data processing. See the canonical, dated sub-processor list; a signed version is part of the DPA.
Report it, and we will respond.
Report a vulnerability
info@araghatta.com. Dedicated address: not published yet; until then use info@araghatta.com PGP: no PGP key is published yet We acknowledge within 2 business days and coordinate disclosure in good faith.
Incident response
Defined runbook; customer notification for confirmed incidents affecting your data within 72 hours, with a root-cause summary to follow.
Product security
Sensor updates are signed and pinned, and staged by your tenant policy. Each release ships a signed software bill of materials.
Service and incident commitments.
What we state today, and the contracted target for each commitment.
| Commitment | What this site states today | Contracted target |
|---|---|---|
| Breach notification | Customers are notified of a confirmed personal-data breach affecting their data without undue delay and within 72 hours of confirmation (DPA), with a root-cause summary to follow. | without undue delay and within 72 hours of confirmation, with a root-cause summary to follow |
| Uptime | Uptime and SLA terms for Discover + Warn and Discover + Warn + Block are set in the order form. | target 99.9% monthly availability for the managed service, not yet independently measured; the contracted commitment is in the order form |
| Recovery (RPO and RTO) | Encrypted database backups with point-in-time recovery; restore drills are planned. | database: Multi-AZ PostgreSQL with point-in-time recovery to within about five minutes and 7-day backup retention (RPO about five minutes). No RTO figure is published yet; recovery objectives are agreed per tier in the order form |
| Incident response times | Defined incident runbook. Vulnerability reports are acknowledged within 2 business days. | severity-based response targets are set per plan in the order form, with 24x7 handling of Severity 1 incidents on the Premier support plan; confirmed incidents affecting your data are notified within 72 hours |
The technical detail, for your reviewers.
Collection schedules, privileges, network destinations, data-handling tables, the sensor deployment guide, the security whitepaper and a recorded product walkthrough are shared with your security and IT team under NDA, with a demo. Request the demo and technical pack.
Related: Security & data handling · Privacy Policy · Terms · DPA · Sensor security whitepaper (PDF)
See the AI on your fleet in about a week.
Run the sensor across your estate and receive an AI risk report and an AI Bill of Materials. The assessment is free and carries no commitment.