SibylMaking Science
Back to Sibyl

Terms of service

Effective September 3, 2026 · Supersedes the version of August 20, 2026

These terms are an agreement between Making Science Group, S.A., C/ López de Hoyos 135, 4º, 28002 Madrid, Spain ("Making Science", "we"), and the person or organization using Sibyl ("you"). If you accept these terms on behalf of an organization, that organization is the contracting party and you confirm that you are authorized to bind it; you do not become a contracting party solely by acting as its representative. These terms cover both parts of Sibyl: the hosted deployment portal at sibylpredicts.com, and the Sibyl application you install into your own Google Cloud project. By accepting these terms at sign-in, or by installing or using Sibyl, you agree to them.

These terms are intended exclusively for organizations and persons acting in the course of their business or professional activity, and are not intended for consumers.

Making Science may retain electronic evidence of the acceptance of these terms, including the date and time of acceptance, the version of the terms accepted, and the information reasonably necessary to identify the person accepting them and the organization on whose behalf they accepted.

What Sibyl is

Sibyl deploys a predictive-audience system into your own Google Cloud project. It trains BigQuery ML propensity models from your Google Analytics 4 BigQuery export, scores your users, writes each user's score band back to your GA4 property, creates the corresponding GA4 audiences, and keeps that loop running on a schedule you control.

The two parts do different things. The hosted portal exists only to confirm you can deploy and to generate a one-line install command; it does not train models, read your analytics event data, or change anything in your Google Cloud or Google Analytics. The deployed application runs entirely inside your project, as a service account in your own IAM, and does all of the work described above there.

Where enabled in your project, two secondary advisory features use Google Vertex AI (Gemini 2.5 Flash) within your own Google Cloud project: one suggests an objective from the goal you type, the other explains a trained model's results. They are intended to assist with configuration and interpretation and do not replace your own professional judgement. You remain responsible for reviewing and validating any AI-assisted output before relying on it or using it for marketing or other decisions.

Sibyl is provided free of charge — your cloud costs are not

Making Science charges nothing for Sibyl. There is no licence fee, subscription, or usage charge, and no payment details are collected.

Everything Sibyl does happens inside your Google Cloud project, so the BigQuery, Cloud Run, and related usage it generates is billed to that project by Google, like any other workload you run there. You are solely responsible for those costs. They depend on the size of your export, how many models you run, and how often you run them — all of which you control. We do not cap, refund, or reimburse them.

What you need, and what you are responsible for

To install Sibyl you need a Google Analytics 4 property where you are an Administrator, a Google Cloud project where you are an Owner with billing enabled, and the ability to run one command. You are responsible for:

  • holding the rights and permissions you assert when you install — including the authority to grant Sibyl access to that property and that project;
  • your own end users: having the privacy disclosures, legal basis, and where required the consent needed to collect their data in Google Analytics and to have it processed and used as Sibyl does, including the association of predictive score bands with them. In respect of that data you act as the data controller and Making Science does not process it;
  • complying with the Google Analytics terms, the Google Cloud terms, and the Google Measurement Protocol and audience policies that apply to your property;
  • how you use Sibyl's outputs — the audiences, scores, and reports it produces, and any marketing or advertising decisions you take on them;
  • the security of your own project: the IAM you grant, the access key that gates your instance, and who you share it with.

Your data stays yours

You keep all rights in your Google Analytics data, your BigQuery data, the models Sibyl trains in your project, and the scores, audiences, and reports it produces there. We claim no ownership of any of it and no licence to use it. It is not sent to Making Science, and it is not used to train anything for anyone else.

What Making Science does receive is limited and described in the privacy policy: your contact details and install record from the portal, and a pseudonymous set of operational events from the deployed instance, which the person who administers it can switch off. The privacy policy governs personal data; where it and these terms differ on that subject, it prevails.

If Making Science separately agrees to provide support or other services that require it to process personal data on your behalf beyond the limited processing described in the privacy policy, the parties will enter into any data processing agreement required by Article 28 GDPR before that processing begins.

Confidentiality

Each party shall treat as confidential any commercial, technical, business, security, or other non-public information disclosed by the other party in connection with Sibyl that, by its nature or the circumstances of its disclosure, should reasonably be regarded as confidential.

Each party shall use such confidential information only for the purposes contemplated by these terms and shall not disclose it to third parties except where necessary for those purposes and subject to appropriate confidentiality obligations, or where disclosure is required by applicable law or a competent authority.

These obligations do not apply to information that is publicly available without breach of these terms, was lawfully known to the receiving party before disclosure, or is lawfully obtained from a third party without an obligation of confidentiality.

Software updates

Sibyl checks for updates and, unless the setting is turned off, applies them to your deployment automatically. This is how security and reliability fixes reach an installed instance, and it may change behaviour or remove features. You can disable automatic updates at install time or in the instance's Settings screen, in which case keeping it current is yours to do. Changes to the SQL a trained model runs are never applied automatically: a deployed model stays pinned to the version it was built on until you adopt an update yourself.

Availability, support, and changes to the product

Sibyl is offered as a free tool, with no service-level commitment. The hosted portal and the release service may be unavailable, interrupted, or discontinued, and we may change, restrict, or withdraw any part of Sibyl at any time. Support is provided at our discretion, by email, with no response-time guarantee. If the release service becomes unavailable, an already-installed instance continues to run in your project; it simply stops receiving updates.

Acceptable use

You must not use Sibyl to:

  • process data you have no right to process, or deploy into a Google Cloud project or Google Analytics property you are not authorized to administer;
  • build audiences that target people on the basis of sensitive characteristics, or in any way Google's advertising and analytics policies or applicable law prohibit;
  • re-identify, or attempt to re-identify, individuals from pseudonymous identifiers;
  • interfere with, overload, probe, or attempt to gain unauthorized access to the portal, the release service, or another organization's instance;
  • resell Sibyl, present it as your own product, or remove or obscure its attribution to Making Science. Renaming your instance to your own organization's name is a supported feature and is not a breach of this.

Intellectual property

Sibyl — its software, container images, SQL, documentation, name, and branding — belongs to Making Science and its licensors. Installing Sibyl gives you a limited, non-exclusive, non-transferable, non-sublicensable, and revocable right to install and run it for your own organization's internal business purposes for as long as these terms are in force. Nothing here transfers ownership, and all rights not expressly granted are reserved. You may not reverse-engineer, decompile, or attempt to derive the source of the container images except to the extent applicable law says you may.

If you send us feedback or suggestions, we may use them freely and without obligation to you.

Google services

Sibyl runs on and integrates with Google services — Google Cloud, BigQuery, and Google Analytics — which are provided by Google under Google's own terms, not ours. We are not responsible for them or for changes Google makes to them. Sibyl's use of information received from Google APIs complies with the Google API Services User Data Policy, including its Limited Use requirements.

No warranty

Sibyl is provided "as is" and "as available", without warranty of any kind, express or implied, including any implied warranty of merchantability, fitness for a particular purpose, accuracy, or non-infringement. In particular, Sibyl produces statistical predictions: we do not warrant that a model will be accurate, that its audiences will perform, or that any commercial outcome will follow. You are responsible for evaluating its output before you act on it. Nothing in this section limits rights you have under mandatory consumer law.

Limitation of liability

To the fullest extent permitted by law, Making Science is not liable for any indirect, incidental, special, or consequential damages, or for lost profits, lost revenue, lost data, or Google Cloud charges you incur, arising out of or relating to Sibyl. Our total aggregate liability arising out of or relating to Sibyl is limited to one hundred euros (€100). Because Sibyl is supplied free of charge, this allocation of risk is a condition of it being offered at all.

Nothing in these terms excludes or limits liability that cannot lawfully be excluded or limited — including liability for death or personal injury caused by negligence, for fraud, or for wilful misconduct.

Ending it

You can stop using Sibyl at any time. The instance's self-destruct removes what Sibyl created in your project; you may also delete the Cloud Run services, BigQuery artifacts, and service account yourself, and revoke Sibyl's Google authorization from your Google Account permissions. Because the application lives in your project, we cannot remove it for you.

We may suspend or end your access to the hosted portal and the release service if you breach these terms or if we discontinue Sibyl. On termination, the sections on confidentiality, intellectual property, no warranty, limitation of liability, and governing law survive to the extent necessary to give them effect. To request deletion of your contact details or install records, write to the address below or see the privacy policy.

Changes to these terms

We may update these terms from time to time, including to reflect changes to Sibyl, applicable law, or regulatory requirements. Where a change materially affects your rights or obligations, we will give reasonable notice before the updated terms take effect — by email to the address you signed in with, or on the portal — unless an earlier change is required by law, for security reasons, or by circumstances outside our reasonable control.

The effective date at the top of this page indicates when the current version took effect. Where acceptance of an updated version is required, we may ask you to accept the updated terms when you next access the hosted portal. If you do not accept an updated version that requires acceptance, you must stop using Sibyl and remove your instance.

Governing law

These terms are governed by the laws of Spain, and the courts of Madrid have exclusive jurisdiction over any dispute arising from them — except where mandatory law in your country of residence gives you the right to bring proceedings elsewhere. If any provision is held unenforceable, the rest remains in force.

Contact

Questions about these terms can be sent to sibyl@group.makingscience.com.