AI Policy

This policy explains how SkillsDB uses artificial intelligence: what our AI features do, how we handle your data, which model providers we work with, and where we expect a human to stay in the loop. It is written for the people who have to evaluate us — procurement, security, privacy, and the HR leaders who will actually use the output.

This page describes how we operate. It is not a contract. Your contractual terms are in your Terms of Service or your signed agreement, and data protection terms are in our Data Processing Addendum.

How we use AI

AI in SkillsDB supports the work of creating, maintaining, and interpreting the information your organization uses to manage skills, roles, and people. Across the product it does five kinds of things:

• Drafts structured content you would otherwise write by hand.
• Extracts meaning from documents, profiles, and free text.
• Compares something against a standard and reports the difference.
• Summarizes patterns across a set of records.
• Suggests what might come next.

Which features use AI changes as the product develops, so we do not maintain that list here. The current features, what each is for, and where each is unreliable are documented in our Documentation.

How we handle your data

• We do not train on your data. We do not use your data to train or improve models made available to any other customer, and we contractually require our model providers not to train on it either.
• We do not pool customer data. Your data is never used to generate output, benchmarks, or comparisons for another customer. Where we show market or comparative information, it comes from publicly available sources, data we license from third parties, and the general knowledge of the underlying models.
• You own what the AI produces. Output generated for you is yours. We assign you any interest we might otherwise have in it.
• Processing location. We currently process AI inputs in the United States. If we add regions, we will say so.
• Working artifacts. Providing AI features involves creating intermediate artifacts such as vector representations and indexes. These exist only to serve you, and on termination they are deleted or rendered permanently non-attributable to you.

Which models we use

We use third-party model providers to deliver AI features. Each one is a subprocessor, and the current list is published in our Trust Center, which is the authoritative source.

We select and change models as the technology moves. That is deliberate: it lets us improve quality and cost without changing what you bought or how you are billed. Where a model change would affect a feature you rely on, we give reasonable notice.

A human stays in the loop

SkillsDB is a tool for people who make decisions about other people. We build it on the assumption that a person, not the software, makes the call.

• SkillsDB does not act as your agent and does not perform employment functions on your behalf.
• Our AI features are not designed to make autonomous employment, compensation, or promotion decisions, and they do not replace your managers or your decision-making processes.
• For any decision with legal or similarly significant effects on an individual, a human being should make, review, or oversee it. Our product is built to support that, not to route around it.
• We do not track how often you accept or override a recommendation, and we do not design toward a target acceptance rate.

How to read AI output

AI output is probabilistic. It can be inaccurate, incomplete, or biased, and it should be reviewed before you rely on it. This is true of every system of this kind, ours included.

Two principles shape how we present it. We describe rather than prescribe: output tells you what the market or your organization looks like, not what you should do. And we favour ranges and distributions over single recommended numbers, because a single number invites more confidence than the underlying data supports.

Where a figure is modeled or inferred rather than directly measured, we work to say so.

What our AI is not for

Some uses are outside what SkillsDB is designed and tested for, and we do not support them:

• Automated screening, scoring, ranking, or rejection of individuals without human review.
• Determining eligibility for employment, credit, housing, insurance, or any similar benefit.
• Inferring protected characteristics, or using output as a proxy for them.
• Legal, compliance, or professional advice of any kind.

Your controls and our obligations

AI features are identified in the product, and administrators control which are enabled. Where a plan includes an allowance of AI activity, usage is visible in the product.

Employment-related AI is regulated differently across jurisdictions, and those rules generally place obligations on the employer using the tool rather than on the vendor supplying it. We publish the feature documentation you need to carry out your own risk, impact, or data protection assessments. Responsibility for notices, opt-outs, record-keeping, and assessments in your jurisdiction remains yours, and we do not warrant that your use of the Services complies with any particular law.

Questions

Security and privacy documentation, including our SOC 2 report and the subprocessor list, is available in our Trust Center. For anything else about how we use AI, contact us at privacy@skillsdb.com.