Governance and trust
Human Command, by design
mAI accelerates marketing work. People keep the decisions that matter: strategy, audiences, budgets, brand-critical creative and regulated claims.
What stays with your people
Every deployment starts from the same rule: mAI recommends and drafts, and a named person decides. The decisions below always have a human gate.
- Strategy changes and new messaging pillars
- Audiences and segmentation, including any use of sensitive attributes
- Budgets above the approved thresholds and major reallocations
- Brand-critical creative and executive content
- Regulated or sensitive claims
- New channels and brand-safety exceptions
- Anything using internal, personal or confidential information
How review is made visible
mAI marks anything that needs a person, in the work itself, with one of four tags. An approval is a dated, written confirmation kept in the approval record. Silence is not approval, and a deadline is never a reason to proceed.
- [Human Approval Required]: the action or content cannot proceed without a named person’s approval.
- [Human Review Required]: a person should check before use.
- [Compliance Check Required]: a compliance or legal review applies, so work pauses.
- [UNVERIFIED]: a fact or claim that could not be supported.
Evidence before claims
Recommendations are matched to the strength of the evidence behind them. A correlation is not treated as a cause. A material budget shift is not recommended from last-click attribution or one platform’s reporting alone. Forecasts are given as ranges with assumptions, and benchmarks are directional only. See the Evidence Methodology.
Named-company models are demonstrations
Models built around named companies use public or non-proprietary context. They are not client engagements and show no results. Measured outcomes come only from real client deployments, anonymized at the client’s request. Browse the Model Gallery and Applied Evidence.
Privacy and data limits
mAI works with the least sensitive data that will do the job, and prefers aggregates and segments to individual records. It does not use or infer protected or sensitive attributes for targeting or personalization unless your regulatory profile and approvals explicitly allow it. Before any outreach is recommended, consent, suppression and do-not-contact status are considered. Write access to your systems is off by default; mAI suggests changes and does not assume permission to make them.
Built to support GDPR, CCPA and PIPEDA obligations. No certification is claimed, and marktgAI does not provide legal advice. Applicable requirements are identified and recorded during scoping. See the Privacy Policy.
AI-assistance disclosure
marktgAI uses AI in its own work, including drafting and analysis. Content on this site that is drafted with AI assistance is reviewed by a person before it is published. mAI models can run on model providers such as OpenAI, Anthropic and Google Gemini, depending on the task, hosting and data requirements of each deployment. Where a law or platform rule requires disclosure that content is AI-assisted, mAI builds that disclosure into the work.
