Security
How Kentro protects your operations data
Kentro is a commerce suite used by people and by AI agents. This page states what is in place, in plain words, and what is not.
- Tenant isolation in the database
- AES-256 encryption at rest
- Two-factor sign-in
- Data stored in the United States
How does Kentro keep each customer's data separate and protected?
Kentro separates each customer's data in the database itself: PostgreSQL row-level security policies bind the tables that hold customer data to the account that owns each row, and the application checks the account again on every request. The production database is encrypted at rest with AES-256 on Google Cloud in the United States. Sign-in supports two-factor authentication.
How your data is kept separate and protected

Tenant isolation in the database
PostgreSQL row-level security policies bind the tables that hold customer data to one account. The services behind the newer app screens, Kentro AI and the MCP server set the account on every request and connect with a database role that cannot bypass the policies: with no account set, a query returns no rows. Every service also checks the account in the application.

Encryption at rest
The production database runs on AlloyDB for PostgreSQL on Google Cloud and is encrypted at rest with AES-256. Google manages and rotates the keys. Database backups are encrypted the same way.

Encryption in transit
Traffic between your browser, your agents and Kentro travels over HTTPS. The production database accepts encrypted connections only. Our Data Processing Addendum lists encryption in transit with industry-standard protocols such as TLS 1.2 or later.

Stored in the United States
Kentro runs on Google Cloud in the United States. The production database and its backups are in the us-central1 region. Vendors that process data for specific features are named, with their locations, on the sub-processor page.
Learn more
Sign-in and access control
Who can get in, and with which rights.
Two-factor authentication
Each user can add a second step at sign-in: a code from an authenticator app, a text message or an email. Kentro staff accounts always require the second step.
Google sign-in
Users can sign in with a Google account. Kentro verifies the Google ID token and matches it to a Kentro user.
Access changes apply on the next request
When an admin deactivates a user or edits their permissions, Kentro clears that user's cached access record, so the change applies on their next request, and within 15 minutes at the latest.
Hashed credentials
Passwords and API key secrets are stored as bcrypt hashes, never as plain text.
Permissions per user
Each user has scopes and permissions that decide what they can see and do. They are resolved on every request, not stored in the token.
Rate limits on sign-in
Sign-in, two-factor and password reset requests are rate limited by client address, with separate limits against password guessing.
How AI and agents reach your data
The questions a careful buyer asks before using Kentro AI or connecting an agent.
Agents sign in as a person
Claude, Codex and other MCP clients connect with OAuth 2.1 sign-in with PKCE and a consent screen in Kentro. The connection is tied to the user who approved it, and its access tokens last one hour.
Kentro MCP serverThe same permissions, never more
Every agent request is checked against the current permissions of that user. An agent sees what the person can see.
Listed and revocable
Agent connections are listed in the app. The person who connected an agent, or an account admin, can revoke it, and a revoked connection is refused on its next request.
Scoped API keys
REST API keys carry the scopes you choose, can have an expiry date and can be switched off. A key pair is sent with each request or exchanged for a bearer token that lasts 15 minutes.
Developers and APIRead-only data questions
When Kentro AI or an agent asks a data question, the query must be a read-only SELECT on an approved list of tables, scoped to your account and capped at 1,000 rows.
No operational writes over MCP
MCP tools read operations data and build reports and dashboards. They do not create orders, labels, stock moves or purchase orders.
What agents can doAI processing and model training
The Kentro AI assistant runs on Anthropic's Claude models, and inference runs in the United States. To answer a question, Kentro sends the prompt and the records the answer needs to the model provider and receives the output.
Our Privacy Policy states: "Kentro's commercial agreements with Model Providers prohibit Model Providers from using Tenant inputs or outputs to train models that are made publicly available." It also states: "Kentro does not use Tenant Personal Data to train foundational AI models for its own account or for third parties."
Kentro may use de-identified, aggregated usage data, such as feature usage, latency and error rates, to run and improve the service. The AI providers we use are named on the sub-processor page.

Audit trail: what is recorded
Kentro records changes where you can read them, and keeps security logs of its own.
- Activity log in Kentro OS: one feed for the whole account, and an Activity tab on orders, invoices, customers, incoming purchase orders and products. Each entry shows the change, the time and the actor.
- Kentro GO keeps its own activity log of policy, campaign and connection changes, with the values before and after.
- Sign-in events: the authentication service records each sign-in attempt with its outcome, IP address and browser.
- Delegated access: when Kentro support staff, or a 3PL provider you work with, sign in to your account on your behalf, they use a separate delegated session that is logged with the identity of the person behind it.
- Infrastructure: administrative changes on Google Cloud are captured in Cloud Audit Logs, and audit logging is enabled on the production database.

Backups and recovery
The production database is backed up automatically every day. Continuous backup is also on, so the database can be restored to a point in time. Backups are encrypted and stored in the same United States region as the database.
We do not publish recovery time or recovery point targets today.

Report a security issue
Email [email protected] with:
- what you found and where: the URL, the API route or the screen
- the steps to reproduce it
- the account you tested with, if any
While you test, please do not open data that is not yours, do not change or delete data, and do not slow down or interrupt the service for other customers.
The same contact is published in /.well-known/security.txt.
Documents for your review
Everything a reviewer needs to attach to a questionnaire.
Sub-processors
The vendors that process customer data for Kentro, what each one does and where.
Learn moreData Processing Addendum
Processing terms, technical and organizational measures (Annex B), breach notification and audit terms.
Learn morePrivacy Policy
What Kentro collects, why, how long it is kept and how AI processing works.
Learn moreTerms and Conditions
The agreement between Kentro and its customers.
Learn moreAcceptable Use Policy
What the service may not be used for.
Learn moresecurity.txt
The machine-readable security contact, in the RFC 9116 format.
Learn more
Security questions
Where is my Kentro data stored?
Customer data is stored in Kentro's production database, AlloyDB for PostgreSQL on Google Cloud, in the us-central1 region in the United States. Database backups stay in the same region. The Kentro AI assistant runs on Anthropic's Claude models with inference in the United States. Other vendors process data for specific features, such as shipping labels or payments, and each one is named on the sub-processor page with its location.
Who can access my data in Kentro?
Your own users, within the permissions you give each of them, and the agents and API keys you connect, within the same limits. Other customers cannot: row-level security in the database and a second check in the application keep each account to its own rows. Kentro staff accounts always require two-step verification. When support staff work inside an account, they use a separate delegated session that is logged with their identity. Our Privacy Policy commits Kentro to role-based access and least privilege.
Is my data used to train AI models?
Our Privacy Policy says Kentro's agreements with its model providers "prohibit Model Providers from using Tenant inputs or outputs to train models that are made publicly available", and that Kentro "does not use Tenant Personal Data to train foundational AI models for its own account or for third parties". Kentro may use de-identified, aggregated usage data, such as latency and error rates, to improve the service.
How do AI agents authenticate to Kentro?
MCP clients such as Claude and Codex use OAuth 2.1: the agent opens a Kentro sign-in and consent screen, a user approves it, and the agent receives a one-hour access token tied to that user. Every request is checked against that user's current permissions. Scripts and integrations use REST API keys with the scopes you choose, sent with each request or exchanged for a 15-minute bearer token. You can revoke a connection or switch off a key at any time.
How do I report a security vulnerability in Kentro?
Email [email protected] with what you found, where you found it and the steps to reproduce it. Please do not open data that is not yours, change or delete data, or disrupt the service while you test. The same contact is published at /.well-known/security.txt on kentro.io, so researchers and scanners find one address.
Does Kentro have a SOC 2 report?
No. Kentro does not hold a SOC 2 report or a HIPAA attestation. What a reviewer can use is on this page: how tenant isolation, encryption, sign-in, agent access and backups work, plus the Data Processing Addendum, which lists our technical and organizational measures in Annex B. Send your questionnaire to [email protected] and we will answer it from these facts.
Does Kentro support single sign-on?
Kentro supports Google sign-in: a user signs in with a Google account, and Kentro verifies the Google ID token and matches it to a Kentro user. There is no SAML single sign-on. Two-factor authentication by authenticator app, text message or email is available on Kentro accounts, and it is always required for Kentro staff.
Questions from your security team?
Send your questionnaire to [email protected], or book a call and bring your reviewer.

