Security and control

Connect securely.
Keep authority.

Review the factual safeguards around Amazon access, data isolation, write controls and the real processing regions used by the platform.

  • Official Amazon OAuth
  • Sealed OAuth profile handoff
  • Preview and approval before writes
Verified safeguards

Security claims should be specific.

These are product controls and documented infrastructure facts—not a blanket claim that data never leaves the EEA.

Official Amazon OAuth

You authorize access on Amazon. Uptova never asks for or stores your Seller Central password.

Scoped OAuth connection

Sealed OAuth handoff

During profile selection, Amazon credentials cross the browser only inside an opaque AES-256-GCM blob bound to the user and returned profiles.

AES-256-GCM · 10-minute replay limit

Account isolation

Row-level security policies restrict application data to the authorized user, account and role scope.

Supabase RLS policies

Controlled writes

Writing engines begin in dry-run. Preflight checks, approvals and decision history keep each supported change traceable.

Dry-run · approval · audit trail

Export and revocation

Activity-tracking data can be inspected, exported and deleted. Disconnecting Amazon removes the stored account tokens.

User data controls · token revocation

No static status claim

Operational status is not presented as live unless it is backed by a real monitored status source.

Evidence before badge

Where processing can occur

Infrastructure locations are stated as documented; international processing is covered by the relevant DPA and Standard Contractual Clauses where applicable.

Supabase
EU · Frankfurt project region
Database and authentication infrastructure
Vercel
EU and US processing may occur
DPA and Standard Contractual Clauses
Cloudflare
Global network
DPA and Standard Contractual Clauses
Product evidence · example data

Applied. Waiting. Blocked.

Every decision leaves evidence. See what changed, what needs approval and what a guardrail stopped. Every value is an example, not a customer result.

No blind writes
Protected write path
Dry-run
Decision logged
Preflight
Scope + guardrails
Approval
Explicit when required
Audit trail
Reason + supported rollback
Decision Board · example data

Every decision leaves evidence.

Illustrative account data
No performance guarantee · No customer data
Decision window
03:47Signals normalized
03:48Decision paths checked
07:30Morning brief ready
Applied automatically

RPC bid bounded

The bid moved inside the approved product scope.

Target TACoS exceeded · RPC −12%−€0.14 bid
Scope and minimum bid passedRollback available
Approval required

Budget increase is waiting

The proposed increase sits above the automatic threshold.

+€80/day proposed€0 moved
Human confirmation requiredPreview ready
Blocked by guardrail

Scale-up refused

A fresh stock signal shows zero available units, so scale-up stays blocked.

Fresh stock quantity: 0€0 moved
Out-of-stock critical blockNo action sent

Dry-run by default

Every writing engine starts in shadow mode: decisions are logged, but nothing is sent to Amazon. Live is always an explicit choice.

Boundaries before approval

Thresholds separate executable actions from decisions waiting for approval or fresher evidence. Unexecuted budget remains visibly at zero.

Protected connection

Amazon connections use the official OAuth flow, account-isolated access and a traceable scope. You never share your Amazon password with Uptova.

See what remains for you after the price.

Your account data · Conservative assumption · No performance guarantee · No write access required for the audit