Skip to content
Log inGet started

AI Policy

DIKA DESIGN LTD, a company registered in England and Wales under company number 16251779, registered office 71-75 Shelton Street, Covent Garden, London, WC2H 9JQ, United Kingdom. Contact: hello@dika.design.

Effective 20 August 2026. We will tell you before a material change takes effect.

This policy is part of the Terms of Service. It explains what happens when you use an AI feature.

1. What happens to your prompt

When you generate text, an image, a video, speech or a transcription, your prompt and any file you attach are sent to the provider that serves the model you chose. You can see which model you are using before you run it.

Infrastructure

  • Amazon Web Services (AWS) hosts the application, the database and stored files, in the European Union (Frankfurt, eu-central-1).
  • Cloudflare provides network protection, DNS and content delivery in front of our services. Requests to us pass through it.

Sign-in

  • Google is offered as a sign-in provider. If you choose it, Google confirms your identity to us and we receive your name, email address and profile picture. We never receive your Google password.

Payments

  • Stripe processes all payments. Card details are entered on Stripe's own pages and are never seen or stored by us.

Email

  • Transactional and marketing email is sent from our own mail infrastructure. We do not use a third-party marketing email platform.

Support

  • Zendesk may handle support conversations depending on how the desk is configured. When it is not in use, support runs entirely on our own systems.

AI providers

Generating text, images, video, speech or transcription sends your prompt, and any file you attach to it, to the provider that serves the model you chose. Which provider that is depends on the model, and you can see the model on screen before you run it. The providers we route to are:

  • OpenAI, Google, Alibaba Cloud (Qwen / DashScope), DeepSeek, fal.ai and OpenRouter for text, image and video generation.
  • ElevenLabs for speech synthesis and voice cloning, and Deepgram for transcription.
  • BytePlus, Runway, Luma, Pika and Kling for video generation.

If you supply your own API key for a provider, that call is made on your own account and under your own contract with that provider. We relay the request and add nothing to it.

A live, always-current list of the providers we actually route to is published at Subprocessors.

2. Your own API key

If you add your own key for a provider, calls made with it are on your account and under your contract with that provider. We relay them and add nothing. Inside a team workspace a personal key is never spent: that is enforced in the product, not a promise.

3. Training

We do not use your content or your prompts to train AI models. Providers operate under their own terms; where we route on our own keys we choose arrangements that do not permit training on customer inputs, and one provider, Runway, grants itself broader rights in its standard terms, which is why we do not route to it on our own keys.

4. Who owns the output

As between you and us, output you generate is yours to use, subject to these terms and the provider's own terms.

Three honest limits:

  • It is not unique. The same prompt can produce the same or similar output for someone else.
  • We cannot warrant it is clear of third-party rights. A generated image can resemble an existing work. Check before you use output commercially.
  • Copyright in purely machine-generated output is unsettled in several countries. You may not be able to register or enforce a copyright in it.

5. It is not advice and it is not checked

AI output is generated, not verified. It can be confidently wrong. Do not rely on it for legal, medical, financial or safety decisions without checking it.

6. Labelling generated media

Where the law requires media that resembles a real person to be identified as artificially generated, that label must stay on it. Do not remove or obscure it. Voice cloning and generated likenesses carry additional obligations under the Acceptable Use Policy.

7. Voice cloning

You may only clone a voice you own or have documented permission to use. We ask you to confirm this and we keep the confirmation, the wording you agreed to, the time and your IP address, as evidence. The training sample is deleted once the clone exists; the consent recording is kept.

8. Cost and limits

AI use is metered. What you spend is shown in the product before and after each run. Plans include an allowance; workspaces can set their own ceilings on automated spending. When an allowance is exhausted we refuse the call and say so, rather than charging you silently.