Responsible automation and client data

Our principles

• Client data remains in client-controlled systems wherever practical.
• Access follows least-privilege.
• No automated external communication is activated without approval.
• Human review remains available for customer-facing output.
• Model providers and data flows are documented before implementation.
• Workflows, credentials, dependencies, and maintenance requirements are handed over.
• The client owns custom workflows built for their business, subject to the agreement.
• Data retention and deletion responsibilities are defined at kickoff.

What this means in practice

Before a workflow is built, we agree which systems hold the data, which accounts the automation uses, and what happens to the output. Anything that reaches a customer starts in review mode and only moves to automatic once the team is satisfied with the quality.

Credentials are created for the client and held by the client. We do not build dependencies that require our continued access to keep a business running.

Handover and ownership

Every workflow ships with documentation covering its purpose, its triggers, its dependencies, and what to do when it fails. The team is trained on it before handover, and ongoing support stays optional rather than structural.