Skip to main content
Digitaltableteur

AI use and transparency statement

Digitaltableteur uses AI-assisted tools in the website assistant and in parts of design, development and editorial work. This statement explains the intended uses, limits, data flow and safeguards. It is a transparency notice, not a legal certification or a promise that every AI output is error-free.

Last updated:

Website AI assistant

Donny is an AI assistant that answers questions about published Digitaltableteur site content, services, work and design-system topics. Replies are generated automatically and are not reviewed by a person before you receive them.

Donny does not rank or profile people, make employment or eligibility decisions, infer emotions or biometric traits, provide legal, medical or financial advice or make binding commitments about price, availability, delivery or contracts. A person must confirm any proposal or commitment.

When you send a message, the text and recent conversation context are sent to the configured model service, using an OpenAI model directly or through Vercel AI Gateway. A temporary browser copy is kept in sessionStorage until you clear it or close the tab. Do not submit sensitive personal data.

AI principles

  • Human accountability: Published work and client deliverables receive substantive human review before publication or delivery. Live website-chat replies are automatic and are not pre-reviewed.
  • Transparency: The AI tools used in each project are documented and disclosed to clients upon request.
  • Privacy first: Personal data is minimized, masked or excluded where the task allows. Training, reuse, retention and transfer settings are checked for the actual provider account and use.
  • Quality assurance: AI is used as an assistive tool, not a replacement for professional expertise and judgment.

Where AI is used

  • Design exploration: generating alternative layouts, color palettes or visual concepts for internal review. All AI suggestions are evaluated and refined by experienced designers before implementation.
  • Editorial assistance: drafting article outlines, alt text descriptions, meta descriptions or translation suggestions. All content is fact-checked, edited for accuracy and approved by a human editor.
  • Code assistance: generating code snippets, documentation or refactoring suggestions using GitHub Copilot or similar tools. All code is reviewed, tested and validated before deployment.
  • Quality assurance: analyzing anonymized support tickets, error logs and telemetry data to identify accessibility issues, performance bottlenecks or user experience improvements.
  • Website chat: Donny uses an OpenAI model, directly or through Vercel AI Gateway, to answer visitor questions from site context. The interface identifies the system as AI before the visitor sends a message.

Where AI is not used

AI is not used for: automated decision-making that affects your rights; high-risk applications as defined by the EU AI Act; final design deliverables without human review; legal, financial or medical advice; or making hiring or business decisions about individuals.

AI providers in use

Provider selection depends on the feature or project. The live website assistant uses an OpenAI model directly or through Vercel AI Gateway. Internal work may also use the tools listed below. Contract, retention, sub-processor and transfer settings must be checked for the actual account and use before personal data is processed:

  • OpenAI: model processing for the website assistant and selected internal drafting tasks. The exact model and routing can change with the deployment configuration.
  • GitHub Copilot: for selected code suggestions and development assistance, subject to the settings and terms of the account in use.
  • Anthropic (Claude): for selected internal reasoning or content-analysis tasks, subject to the settings and terms of the account in use.

A provider name alone is not proof of compliance. Digitaltableteur reviews the relevant service terms, data-processing terms, transfer mechanism, sub-processors, data-use controls and retention settings for the intended use.

Data sent to AI providers

Strict data minimization principles apply:

  • Only necessary prompts: only the minimum information required for the specific task is shared
  • Personal data minimized: client workflows remove or mask identifiers where the task allows; website-chat text is sent as entered, so visitors are told not to submit sensitive personal data
  • Training and reuse controls: the applicable account settings and provider terms must prevent submitted client or visitor data from being used for model training unless a separate lawful, disclosed decision permits it
  • Sensitive data excluded: confidential business information, credentials and special-category personal data must not be entered into public or website AI tools

For website chat, Digitaltableteur does not write the transcript to an application database in the chat route. The browser keeps a temporary sessionStorage copy until you clear it or close the tab. The configured model service processes the submitted text and may keep service or abuse-monitoring records under the applicable account terms.

Governance and safeguards

  • Human review: published work and client deliverables are reviewed for substance, accuracy, sources, bias and appropriateness. Live assistant replies are instead bounded by interface warnings, system rules, source-grounding controls and a route to report a problem.
  • Data minimization: in client work, identifiers are removed or masked where the task allows. Website chat is sent to the model service as entered, with only recent turns included, which is why the chat asks you not to share sensitive personal data. Technical and organizational measures protect what is processed.
  • Vendor review: only providers that offer EU-grade security, data processing agreements (DPAs) and compliance with GDPR and emerging AI regulations are used.
  • Monitoring and auditing: logs of AI usage in client projects are maintained and periodic reviews are conducted to ensure compliance with Digitaltableteur’s policies and standards.
  • Team training: Digitaltableteur staff receives guidance on responsible AI use, prompt engineering and identifying AI limitations or biases.

Limitations and accuracy

AI systems can produce inaccurate, incomplete or biased outputs and can miss context. Donny’s live replies are not professional advice and are not reviewed before display. Verify important information using the linked site source or contact Petri before relying on it. Published work and client deliverables follow a separate human editorial review.

Your rights

  • Right to information: you can ask Digitaltableteur which AI systems were used in your project or to process your data at any time. A clear explanation will be provided.
  • Right to object: you may request that AI-assisted tools are avoided in your project. Feasibility will be confirmed and the alternative workflow documented. Some projects may require additional time or cost adjustments.
  • Right to erasure: if AI tools processed personal data related to you, you can request deletion. Data will be erased in accordance with Digitaltableteur’s retention policies or contractual obligations.
  • Right to human review: if you believe an AI-generated output adversely affected you, you can request human review and challenge the outcome.

To exercise these rights, email mail@digitaltableteur.com or use the channels listed in the privacy policy. Digitaltableteur will respond within one month.

Changes to this statement

As AI technology and regulations evolve, this statement may be updated. Changes will be posted on this page with an updated date. Clients and visitors are encouraged to review this statement periodically.

Contact

For questions, requests or concerns about this AI use and transparency statement, contact: mail@digitaltableteur.com or +358 45 657 4469