Privacy & data use.
Know what happens to your work, and how it may be used to train AI.
Public website and deployed service are separate. Demo tasks stay in your browser; deployment and any research data use follow the applicable environment-specific agreement.
Access and deployment requests
When you register or request deployment, Koyalab saves your email address, language, contact-request time and privacy-notice version in a Cloudflare D1 database. Our team uses these details to coordinate invitations and deployment, not for model training or unrelated marketing. Your email is not verified by this form. Registration adds you to the contact waitlist; it does not activate an account, establish a connection or grant model access. No password, prompt or task content is collected by the registration form.
Contact details are kept while your request is pending, until you withdraw it or coordination is completed. To withdraw or request correction/deletion, contact hello@koyalab.xyz from the registered email address. Rate limiting temporarily stores a salted network-address hash and counters, not the raw address in the waitlist database. Hosting providers may process network information separately. Provider backups may retain deleted copies under their retention schedule; immediate erasure of all backups is not promised.
Free compute credit applications
Individuals and authorized organizations can request research-sponsored compute: support in exchange for explicitly licensed task data for model training and evaluation. Before any such award, our team confirms eligibility, authorized sources, data scope, recipients, permitted uses, retention, safeguards and withdrawal limits in a separate agreement. No company, client or third-party data may be included without the necessary authority. The application itself does not record that research authorization.
To apply for free credits to use our compute, the application form collects your name, email, optional organization, task/use-case summary, compute needs, language, request time and notice version. These are stored in our Cloudflare D1 database for our team to assess the request and contact you about eligibility, any award and deployment. This is not a credit purchase, guaranteed award, account activation, compute reservation or permission for model training. No payment details, credentials, private datasets or task files are requested.
Application details are kept while assessment and access coordination are pending, until you withdraw or coordination is completed. Contact hello@koyalab.xyz from your registered address for correction, withdrawal or deletion. Repeated submissions from the same normalized email keep the first application; email us to update it. The form does not send an automatic email or allocate credits. Rate-limit and backup handling are as described for contact registration above.
Public demo data
This website includes a public demo, separate from deployed Kinetix workspaces. Tasks you write, project and branch labels, sample playback progress, review markers and language preference stay in this browser on this device. This demo does not send task content to Kinetix, a model provider or a research lab. Sample results are illustrative, not live model output.
Deployed workspaces
You can request assisted deployment using your compute, model/API credits and enterprise environment. Our team confirms the supported scope, access boundaries, providers and data handling before setup. Submitting a request does not create a workspace or execute a task.
Before processing workspace data, the deployment notice and agreement must identify the contracting/data-handling entity, authorized administrators, providers and recipients, storage and retention, backup deletion, transfer arrangements and available access, correction, withdrawal and deletion procedures. This website does not replace those deployment-specific terms. Your old local tasks are never uploaded automatically.
Model training and research use
Model training and evaluation, including sharing with partner AI research labs, are separate purposes from completing your task. Any such use requires the applicable authorization and agreed data scope. When you explicitly agree to the research-data terms presented for your participation, the authorized data may be used for model training and evaluation within that agreement. Only material you have the right to license may be included; existing applications and private workspace data are not automatically included.
An agreed research scope may cover task instructions and conversations, code or document context, agent actions, tool inputs and outputs, errors, retries, final results and feedback. The applicable notice must specify the purpose, data categories, authorized sources, recipients, retention, filtering and de-identification measures, and withdrawal or deletion limits before collection or sharing. Do not assume that a deployment is opted into training or that confidential enterprise data is eligible.
This public demo does not collect or share task content for training. Trying a sample does not authorize an upload of your local tasks. Research participation is arranged through an explicit data agreement before collection or sharing. Application contact details are used for coordination, and browser-local demo tasks are not uploaded to research recipients. Withdrawal cannot guarantee removal from models already trained or independent third-party copies; applicable limits must be explained in the agreement.
Permissions and sensitive information
Do not enter passwords, API keys, private credentials or content you do not have permission to share into this website. For workplace or client projects, confirm confidentiality, administrator authority and data-use requirements before deployment. The public demo does not automatically remove sensitive information from text you save or download. Any deployment-specific safeguards must be agreed and validated; no automatic-redaction guarantee is made here.
Access, downloads and deletion
For the public demo, you can review or delete tasks, download a workspace backup or clear local data in Settings. Task content stays until you delete it or clear browser storage. Downloads are separate copies; clearing local data does not delete those files. The demo has no cloud task synchronization. For a deployed workspace, use the controls and contact procedures specified for that environment.
Website delivery
The website is delivered through Cloudflare and loads fonts from Google Fonts. Your browser sends the network and request information needed to load those resources, such as IP address and browser information, to those services. We have not added advertising trackers in the application code. Hosting-level request measurement is separate from collection of task content.
Contact and updates
Kinetix is a product by Koyalab. For deployment, data-use questions or rights requests, contact hello@koyalab.xyz. This notice distinguishes the public website and demo from deployed workspaces and separately authorized research work. Material changes to processing require the applicable notice and authorization before they take effect.