The model lives in the box
Ask Phynix and the AI Brain run on the appliance. No data goes to an outside AI by default.
Home / Security & trust
Security & trust
A security product should be secure first.
Phynix runs inside your network, keeps its AI on the appliance, protects its own console and gives your auditors the evidence.
Your data stays yours
Nothing leaves unless you allow it.
Allowlist only
New installations talk only to your network and to internet hosts an administrator has explicitly allowed.
Built for DPDP
E-mail bodies are never stored; secrets are removed from documents before indexing; the AI sees only what each user may see.
Phynix Guard
A guardian that guards itself.
Ten failed sign-ins, blocked
Password guessing or probing of the console blocks the address for 30 minutes and opens an incident.
Only your admin networks
Limit the console to the subnets you list. Phynix refuses a setting that would lock you out.
Checked every 15 minutes
Phynix fingerprints its own program files and host; any change outside an update raises an incident.
Every weak setting, with the fix
Two-factor, admins, certificate, SSH, firewall, key age, backups and egress, scored PASS, WARN or FAIL.
Cannot be talked into it
Logs, documents and e-mails are treated as data, never as instructions. Answers are scrubbed of secrets.
Risky services flagged
Telnet, exposed databases, the Docker API, remote desktop on user PCs and expiring certificates, alerted without touching your systems.
Controls
What your security team will ask.
| Question | Phynix |
|---|---|
| Who can sign in? | Named users with roles and site-limited access; two-factor sign-in built in and enforceable by policy. |
| What accounts does it use on my servers? | A read-only account to discover and watch, and a restart-only account for named services. Neither is an administrator. |
| Where are credentials kept? | Encrypted on the appliance and never shown again, or fetched at use from HashiCorp Vault, CyberArk or BeyondTrust. The encryption key can be rotated. |
| Can it change things on its own? | Only safe service restarts within limits. Critical servers ask first; a kill switch stops everything; four-eyes approval is available. |
| Is there an audit trail? | Every plan, approval, action, check and undo is recorded in a tamper-evident trail. |
| Can we use our own certificate? | Yes, from your certificate authority, with automatic rollback if it fails. |
| What software is inside? | A software bill of materials (CycloneDX) on request; libraries audited for known vulnerabilities each release. |
—
Days until the DPDP Rules' core duties apply on 13 May 2027. Phynix measures the technical safeguards automatically (access control, encryption, logging, backups, patching, incident readiness) and lists the policy items your team signs off.
Talk to us
See it on your own network.
A 2-day health check, then a 15-day live pilot. Write to us or book a live demo.
office@phynix.co.in