Control over data, access and every action.

• Data handling

  • Read broadly, write narrowly: Procesario reads event data from your systems and data platform; it writes back only through typed, permissioned, logged actions.
  • Designed to compute against the customer's data platform where possible rather than duplicating it into another warehouse.
  • Individual identifiers are intended to be pseudonymized by default in analytics; the platform models work, not workers.

• Identity and access

  • Role-based access with attribute-based data scopes for people.
  • Agents and automations act under their own identities with explicit scopes, like service accounts.
  • Enterprise single sign-on and user provisioning are planned for the productized platform.

• Governed actions

  • Every action type runs at an explicit autonomy level (A0–A5) with an approval requirement, value and volume limits and a reversibility class.
  • Policies such as approval limits and segregation of duties are machine-checkable rules.
  • A kill switch stops execution globally, per playbook or per executor.

• Audit and evidence

  • An append-only record of every prediction, recommendation, decision, action and result, linked to its case.
  • Human decisions are logged for accountability of actions, not for evaluation of people; access to per-user decision logs is restricted to audit roles.
  • Export of the record to the customer's own systems is part of the design intent — the record is the customer's data.

• AI model use

  • Model providers are interchangeable services; bring your own keys.
  • Agent steps receive bounded context and a permitted tool set, and produce a schema-checked output.
  • Cost budgets per playbook and executor are planned; model versions and calibration are tracked.

• Certifications

  • Procesario does not currently claim SOC 2, ISO 27001 or ISO/IEC 42001 certification. Independent assurance is part of the productization roadmap and will be stated here only when achieved.
  • Security questionnaires can be requested through the contact page.

This page describes design principles and roadmap intent. Contractual security terms are agreed per engagement. For a security review, request a conversation and mention security in your message.