how we work
Ethical, privacy-first AI.
We treat your data, your customers, and your brand like our own.
Nothing auto-sendsAI output remains a draft until you approve it.
Only data you approveWe use only the information you authorize for the project.
Responsible data handlingWe agree on data retention and deletion practices before work begins.
You own the workThe deliverables we create for you are yours, as defined in our agreement.
01
Data handling
- We only use data you explicitly approve for the project and its intended purpose.
- We minimize sensitive information and avoid putting it into public AI tools.
- For demonstrations and testing, we use de-identified or synthetic data whenever feasible.
02
Tool selection
- If your company prohibits a particular tool, we discuss suitable alternatives.
- Whenever feasible, we work within your existing accounts and tools so you remain in control.
03
Access and security
- We request only the minimum access and permissions required to complete the work.
04
Data lifecycle and deletion
- We agree upfront on what information we receive, where it will be stored, who can access it, and how long it will be retained.
- At project close, we delete local copies within the agreed timeframe and confirm completion in writing.
- We discuss any applicable backup or retention limitations in advance.
05
Transparency and approvals
- We document delivered workflows in plain English, including their purpose, steps, tools, and relevant privacy considerations.
- External integrations, automations, or data connections require your approval.
- We work within your company's legal and IT policies and adjust our approach based on your feedback.
06
Human oversight and safety
- AI outputs remain drafts until a person reviews and approves them. Nothing is automatically sent to your customers without your authorization.
- We review and test AI outputs for potential errors, including inaccurate information, before recommending production use.
- We review results with you and then refine prompts, information sources, or workflows as needed within the scope of work.
07
Copyright and intellectual property
- We avoid using third-party content that you do not own or have permission to use.
- Deliverables, custom AI helpers, prompts, and workflow documents created for you belong to you, as defined in our agreement.
- We do not reuse your proprietary datasets or workflows in other client engagements.
- We do not share confidential competitive strategies between clients without explicit permission.
08
Marketing use and confidentiality
- We do not use your name, logo, or results in marketing materials without your written permission.
- Testimonials and case studies are submitted for your approval and can be anonymized upon request.
09
Incident response
- If we identify a data exposure or security issue, we promptly notify your designated contact and explain the known scope and recommended next steps.
- We pause affected workflows until the identified risks are addressed and document the resolution.
10
Compliance alignment
- Marsten AI is not a law firm. We follow your legal counsel's guidance when appropriate.
- If your project involves regulated or particularly sensitive information, we discuss the necessary safeguards and agreements before proceeding.
- If your organization has specific data-residency requirements, we discuss those requirements in advance.
11
Your choices and controls
- You determine your data-residency preferences where supported by the selected tools.
- You approve the information sources an AI workflow may access and the channels through which it may communicate.
Still have questions?
Email us at hello@marsten.ai