Notice to users
You are interacting with an artificial intelligence system.
Citead analyses and scores web content and produces recommendations and texts by means of artificial intelligence models.
The results are generated automatically. They may contain inaccuracies, approximations or wrong references and must be checked before any use or publication. They do not constitute professional advice.
Editorial content produced by Citead is marked as generated by an artificial intelligence.
This notice is shown in the solution's interface, in accordance with article 50(1) and (5) of the AI Act. This page is the full notice it refers to.
Identification of the provider
| Item | Information |
|---|---|
| Provider | TW3 PARTNERS, simplified joint-stock company |
| Registered office | 45 rue du Cardinal Lemoine, 75005 Paris, France |
| Registration | RCS Paris 949 610 570 |
| Legal representative | Angela Procoli, Managing Director |
| "Artificial intelligence" contact | contact@tw3partners.com |
| AI system concerned | Citead (formerly DataGeo) |
| Version | 1.0. The system is delivered continuously; this notice is updated at each significant change of the features or the models |
| Date of placing on the market | 27 January 2026, date the first public analysis tool went online. Paid subscriptions are not yet open |
Qualification under the AI Act
TW3 PARTNERS acts as provider within the meaning of article 3, point (3), of Regulation (EU) 2024/1689: it develops an artificial intelligence system and places it on the European Union market under its own name and trademark.
The professional customer who uses Citead under its own authority acts as deployer within the meaning of article 3, point (4), of the same regulation, and as such bears the obligations recalled below.
TW3 also acts as downstream provider within the meaning of article 3, point (68), in that it integrates into its solution general-purpose artificial intelligence models developed and made available by third parties.
Classification by level of risk
| Category | Applicable to Citead |
|---|---|
| Prohibited practices (article 5) | No. Citead uses none of the prohibited practices, in particular no subliminal or manipulative technique, no exploitation of vulnerabilities, no social scoring, no emotion recognition in the workplace or in education, no biometric categorisation and no remote biometric identification. |
| High-risk system (article 6 and Annex III) | No. Citead is not intended to be used in any of the areas listed in Annex III and is not a safety component of a product covered by Annex I. |
| Transparency obligations (article 50) | Yes. Citead interacts directly with natural persons and produces synthetic content. The corresponding obligations are described below. |
| General-purpose AI model (Chapter V) | TW3 is not the provider of a general-purpose AI model. It integrates models supplied by third parties, which remain subject, on their own account, to the obligations of Chapter V. |
Description of the system
Intended purpose. Citead is a professional tool for analysing content and assisting its production. It is intended to be used by marketing, communication, publishing and search professionals, in the course of their business.
Main features:
- analysis of web pages and content and assignment of scores against performance criteria for search engine optimisation (SEO) and visibility with generative answer engines (GEO);
- production of optimisation recommendations;
- assistance with the drafting and generation of content intended for publication.
Operation. From the content, page addresses and instructions supplied by the user, the solution runs proprietary analysis processing and sends requests to general-purpose artificial intelligence models to produce scores, analyses and texts.
Artificial intelligence models used:
| Model | Provider | Access channel | Role in the solution |
|---|---|---|---|
| DeepSeek V4 Flash | DeepSeek, served by Scaleway | Scaleway Generative APIs, fr-par region (Paris) | Page analysis, GEO scoring, keyword extraction, classification, generation of titles and articles, assisted rewriting |
| bge-multilingual-gemma2 | BAAI, served by Scaleway | Scaleway Generative APIs, fr-par region (Paris) | Search in the knowledge base |
| Whisper large v3 | OpenAI (open model), served by Scaleway | Scaleway Generative APIs, fr-par region (Paris) | Audio transcription |
| Gemini 2.5 Flash, Gemini 3 Flash, Gemini 2.5 Flash Image | Google programming interface | Documentary research with Google Search, video analysis, image generation | |
| Claude | Anthropic | Anthropic programming interface, in the United States | Drafting of the free pack sent after a page analysis: article topics, rewriting of the page, draft articles |
The provider and version of the models change: the list above describes the configuration in force on the date of this notice. The models served by Scaleway run in France. Processing entrusted to Google and Anthropic runs under the conditions described in Annex 3 of the data processing agreement.
TW3 reserves the right to change the models used and their versions. This notice is updated accordingly.
Infrastructure. The solution is hosted by Scaleway SAS in France, in the fr-par region (Paris).
Data used. The solution processes the content submitted by the user as well as the publicly accessible data of the web pages whose analysis is requested. Content submitted by customers is not used to train, retrain or fine-tune the models, whether TW3's models or those of its providers.
Known limits and risks
Known limits. The customer is informed that:
- the content produced may contain factual inaccuracies, unverified statements and non-existent references, including when it is worded confidently;
- the scores and recommendations rest on statistical models and heuristic criteria, and are neither a guarantee of result nor professional advice, in particular legal, financial, medical or tax advice;
- the results may vary over time with changes in the underlying models and in the algorithms of search engines and generative answer engines;
- the models may reproduce biases present in their training data;
- similar content may be produced for different users from comparable instructions;
- the content produced may, in some cases, show similarities with protected works, and TW3 cannot guarantee that it does not infringe.
Uses not intended. Citead is not designed and must not be used:
- to take decisions producing legal effects concerning natural persons or similarly significantly affecting them;
- in any of the areas covered by Annex III of the AI Act, in particular employment, education, access to essential services, credit, law enforcement or justice;
- to produce content intended to mislead the public about its artificial origin;
- to produce disparaging, defamatory, discriminatory or misleading content, or content amounting to unfair commercial practices;
- to process sensitive data within the meaning of article 9 of the GDPR or data concerning minors.
Human oversight
Citead is designed as a tool to assist decision and production. It takes no decision on its own and publishes nothing without human intervention.
The customer must organise an effective human review of any content produced before using, distributing or publishing it. This review must be entrusted to people with the skills, training and authority needed to:
- check the factual accuracy of the statements and the existence of the sources cited;
- assess the lawfulness of the content, in particular with regard to the rights of third parties;
- discard or correct any result that is manifestly wrong, biased or inappropriate;
- decide, as a last resort, on publication.
Transparency obligations borne by TW3
Information about interaction with an AI system (article 50(1)). As Citead is intended to interact directly with natural persons, TW3 informs them, at first contact, that they are interacting with an artificial intelligence system. This information is given by the notice reproduced at the top of this page, shown clearly and distinguishably in the interface.
Marking of synthetic content (article 50(2)). Editorial content produced by Citead is synthetic content within the meaning of article 50(2) of the AI Act.
State of implementation: each piece of generated content carries, in the solution's database and in the responses of the programming interface, the identification of the model that produced it and the timestamp of its production. Machine-readable marking of the content itself, intended to make it detectable after export or publication, is not yet deployed. Its implementation is under way; this notice will be updated when it is effective, with a description of the method chosen.
The customer and its users undertake not to remove, alter, conceal or circumvent the marking elements in place.
Clarity of the information (article 50(5)). This information is provided at the latest at the time of the first interaction or exposure, clearly and distinguishably, and in compliance with the applicable accessibility requirements.
Obligations of the deploying customer
The customer who uses Citead under its own authority bears the following obligations, which TW3 cannot fulfil in its place.
AI literacy (article 4). The customer takes the measures needed to ensure, to the best extent possible, a sufficient level of AI literacy of its staff and of the people dealing, on its behalf, with the operation and use of the solution, taking into account their technical knowledge, experience, education and training and the context of use. This obligation has applied since 2 February 2025.
Information on the artificial origin of published texts (article 50(4)). The deployer of an artificial intelligence system that generates or manipulates text published with the purpose of informing the public on matters of public interest must disclose that the text has been artificially generated or manipulated. This obligation does not apply where the AI-generated content has undergone a process of human review or editorial control and where a natural or legal person holds editorial responsibility for its publication.
In practice, it is for the customer:
- to identify, among its publications made with Citead, those that deal with matters of public interest;
- either to state their artificial origin clearly and visibly;
- or to organise and document a process of human editorial control that allows it to rely on the exception, and to identify the person who holds editorial responsibility.
TW3 recommends, in every case, putting such a documented editorial control process in place.
Use in line with the intended purpose. The customer uses the solution in line with its intended purpose and with this notice. It refrains from any use excluded under the known limits and risks.
Reporting. The customer informs TW3 without delay of any malfunction, any manifestly wrong or harmful result, and any incident it becomes aware of, at contact@tw3partners.com.
Point of attention: change of qualification (article 25). The customer is informed that it could be regarded as provider of the artificial intelligence system, and bear all the corresponding obligations, if it were:
- to put its name or trademark on the solution;
- to make a substantial modification to it;
- to change its intended purpose in such a way that the system becomes a high-risk system.
Such conduct is prohibited by the terms.
Governance, cooperation and contact
Monitoring and updating. TW3 monitors the behaviour of the solution and updates this notice at each significant change of its features, of the models used or of the applicable regulatory framework.
Cooperation with the authorities. TW3 cooperates with the competent national authorities and gives them, on a reasoned request, the information needed to demonstrate the compliance of the solution.
Complaints. Anyone may send a complaint about the use of the solution to contact@tw3partners.com. TW3 acknowledges receipt and replies within a reasonable time.
Related documents: terms of sale and use, data processing agreement, privacy policy, security measures (article 32 of the GDPR).