Kadentra / Trust center

Every change to your Salesforce orgneeds a person’s approval.

Kadentra turns a plain-language request into a tested Salesforce change. It works in an isolated environment, records every step, and stops twice for a person to decide. One person confirms the requirement before work starts, and one approves the release before anything ships.

Last updated November 2026

Principles

Governance is the product, not a feature.

Every part of Kadentra starts from one question: who is accountable for this decision? The agent does the work in between. People make the calls.

01

You approve the requirement

Before anything is built, Kadentra shows you what it understood and where that came from in your org. Nothing moves until you confirm it.

02

You approve the release

You see the test results and the evidence for each change. A person signs off. The agent never releases its own work.

03

Your production org stays untouched

Building and testing happen in an isolated Salesforce environment, never in production.

04

Every step is written down

Every requirement, story, test and build step links back to the original request.

How a request flows

Nothing ships until you say so.

  1. Business request
    Plain language
  2. Refined requirement
    Gaps surfaced
  3. Gate 1
    Right thing to build?
  4. User stories
    Acceptance criteria
  5. Test cases
    Before the build
  6. Build
    Isolated environment
  7. Test
    Evidence recorded
  8. Gate 2
    Safe to ship?
  9. Release package
    Yours to deploy
Agent step
You decide

Plain-language gates

You review what will change in business terms.

Review by exception

Work between the gates is checked against what you approved. Anything that drifts from it is flagged to a person.

Waits as long as you need

Approve hours or days later and the work picks up where it stopped.

Your data

Built so your data stays where it belongs.

Not used to train models

Your requests, org metadata and outputs are never used to train AI models.

Isolated per customer

Each customer's data is separated at the database level, not only in application code.

Encrypted everywhere

Data is encrypted at rest and in transit.

Kept out of our logs

Content from your org is redacted before it can reach logs or error messages.

Yours on request

Users can export their data at any time, and we handle erasure requests.

Hosted in the United States

Kadentra runs on AWS in the United States.

Layers of protection

Five layers between a request and your org.

  1. Layer 01

    Secure sign-in

    Managed identity with session limits and hardened cookies.

  2. Layer 02

    Tenant isolation

    Database-enforced separation between customers.

  3. Layer 03

    Isolated build environment

    Every requirement is built and tested away from your production org.

  4. Layer 04

    Human gates

    Two approval points that pause the work until a person decides.

  5. Layer 05

    Tamper-resistant audit trail

    Every change, and every access by our staff, is recorded and can't be edited afterwards.

AI safety

Security made for how AI agents fail.

Grounded in your org

The agent works from your actual metadata, not guesses.

Bounded actions

The agent can only carry out a fixed set of Salesforce operations, inside the isolated environment.

Built-in limits

Every stage has a limit, so an agent can't run unchecked.

Checked against what you approved

Any deviation from the approved requirement is flagged.

Every decision traceable

Agent steps are recorded so they can be reviewed later.

Operations

Watched around the clock.

Continuous threat detection

With alerts to our team.

Full infrastructure audit logging

Hardened web and API layer

Least-privilege access

For systems and staff.

Defined data retention

Operational data is deleted on a fixed schedule.

Subprocessors

The providers that run Kadentra.

We use a small set of trusted providers to run Kadentra.

View our subprocessor list

FAQ

Questions security teams ask us.