Acceptable Use Policy

Intended purpose

Sentia is intended to help product teams explore how described segments of people might respond to a product decision, before that decision is made. It is designed for exploratory and directional research, conducted by professionals who bring their own judgment to the result.

Decisions about people

You may not use Sentia, or any output from it, to make or materially inform a decision about a specific individual. This is the most important restriction in this policy and we do not grant exceptions to it.

Specifically, you may not use the platform or its output in connection with:

• employment decisions of any kind, including hiring, screening, work allocation, evaluation, compensation, promotion, discipline, or termination;
• credit, lending, or any decision about a person's access to or pricing of financial services;
• insurance eligibility, underwriting, or pricing;
• housing, tenancy, or real-estate decisions;
• education admissions, placement, assessment, or accreditation;
• healthcare decisions, clinical judgment, diagnosis, treatment, or eligibility for care;
• legal judgments, immigration or asylum determinations, law enforcement, or the allocation of public benefits or essential government services;
• social scoring, or evaluating or classifying people on the basis of their social behavior, personal characteristics, or biometric data; and
• any use where a failure of the output could reasonably result in death, personal injury, or severe environmental damage.

You may not use the platform to infer sensitive characteristics about identifiable individuals, including health status, disability, racial or ethnic origin, religious or philosophical belief, trade union membership, sexual orientation, or sex life. You may not use it in any way that discriminates against a person or a group on a protected ground.

Presenting results

Simulated results carry an obligation about how they are described. When you share a Sentia result with anyone, inside or outside your organization, you must not present it as having come from human participants.

You may not:

• describe simulated agents as participants, respondents, panelists, interviewees, or customers, or otherwise imply that a real person gave the response;
• describe a simulation as a poll, a survey, a census, or a study of public opinion;
• attach a margin of error, a confidence interval derived from sampling theory, or any other statistic that implies a random sample of humans, to a simulated result;
• remove, obscure, or alter any label, marker, or provenance metadata we apply to identify output as AI-generated, including in exports and shared links; or
• combine simulated results with data collected from human participants without clearly identifying which is which.

If you belong to a professional research body, its code will require you to disclose the use of AI and synthetic data in your reporting. The platform records methodology and limitations with each completed report.

Advertising and public claims

You may not use simulated results to substantiate a public claim about what real people think, prefer, or do.

This means you may not use output from Sentia as, or to generate, a consumer review, a testimonial, or an endorsement, and you may not use it to support a claim of the form “most users prefer” or “nine out of ten customers say”, in advertising, marketing, packaging, investor materials, regulatory submissions, or any other public communication. A claim of that kind represents that a real person held the opinion. A simulated agent is not a real person and did not hold it.

You may of course use Sentia to decide what to test, to draft a claim you then validate with human research, and to publish results that are clearly described as simulated.

Twins of real people

Where the platform supports building an agent grounded in a specific, identifiable person's interview, additional rules apply, and they are not waivable.

• You must have that person's informed consent, recorded before the agent is created, covering both the creation of the agent and its use in simulations. Consent to be interviewed is not consent to be simulated.
• You must honor a withdrawal of consent. Withdrawal erases the agent, its embeddings, and its derived memory.
• You may not use such an agent to generate statements attributed to that person outside your own workspace, to endorse anything, or in any public-facing or promotional material.
• You may not create an agent modeled on a real person without their consent, including a public figure, and you may not use the platform to impersonate anyone.

Data you bring

You are responsible for the research materials, transcripts, recordings, analytics data, and integration content you bring into the platform. You must have the rights and the consents needed for us to process it on your behalf, including any consent required from the individuals the data is about.

You may not upload or connect:

• personal information about anyone under 13, or about anyone under 18 without the consents applicable law requires;
• data you are contractually or legally barred from disclosing to a processor, or that you obtained unlawfully;
• protected health information, payment card data, or government identifiers, unless we have agreed in writing in advance that your workspace may process them; or
• credentials for systems you are not authorized to access. Authenticated research runs use the approved capture flow, and inline raw credentials are rejected.

Platform integrity

You may not:

• attempt to gain unauthorized access to the platform, another customer's workspace, or any underlying system, or probe or test our security other than as described on our Security page;
• interfere with or disrupt the service, including by circumventing rate limits, usage caps, credit accounting, or entitlement checks;
• use the platform to develop a competing product, or to benchmark it for publication, without our written consent;
• reverse engineer the platform, or attempt to extract the underlying models, prompts, or weights;
• use the platform to direct our infrastructure at systems you do not own or have permission to access, including through browser research runs; or
• use the platform to generate or distribute malware, spam, or unlawful, harassing, or infringing content.

How we enforce this

If we believe this policy has been breached, we may suspend or limit access to the affected workspace, remove or disable the material involved, or terminate the agreement under the Terms of Service. Where the breach is capable of remedy and does not present an immediate risk to other customers, to individuals, or to the platform, we will give you notice and a reasonable opportunity to fix it before suspending.

Where we can act more narrowly than a suspension, we will. Where a breach involves a risk of harm to an identifiable person, we will act immediately.

To report a suspected breach of this policy, write to hello@sentialabs.ai. We may update this policy as the product and the law change, and we will give notice of material changes as described in the Terms of Service.