![questel_logo](/questel_logo.png)
- PatentNew
- Trademark
- Innovation
- Solutions
- Contact
- Learn & Support
- Learn and support
- Resource HubAccess value added content to support your IP strategy
- Webinars & EventsAre you interested in attending one of our online or onsite event?
- Product TrainingsCustomer success is our priority. Increase your skills in the use of Questel’s software
- Product NewsA platform dedicated to software and platforms news and evolutions
- Best-in-class Customer ExperienceOur goal is to exceed our clients' expectations and share best practices
- IP TrainingIncrease the IP-IQ of your entire organization with engaging IP training programs
- Resource Hub
- About Questel
- Learn & Support
- Learn and support
- Resource HubAccess value added content to support your IP strategy
- Webinars & EventsAre you interested in attending one of our online or onsite event?
- Product TrainingsCustomer success is our priority. Increase your skills in the use of Questel’s software
- Product NewsA platform dedicated to software and platforms news and evolutions
- Best-in-class Customer ExperienceOur goal is to exceed our clients' expectations and share best practices
- IP TrainingIncrease the IP-IQ of your entire organization with engaging IP training programs
- Resource Hub
- About Questel
- Home/
- Trusted AI
Questel Deploys AI
You Can Trust
Questel leverages proprietary Generative AI models from OpenAI through API integrations (OpenAI API Platform), enhanced by prompts optimized by our in-house experts to ensure peak performance for our clients. We also utilize open-source models like Mixtral, Falcon, and Llama 2, which we fine-tune and train on our proprietary data sets. This approach enables us to develop advanced features for exploring intellectual property information, providing cutting-edge solutions tailored to your needs.
![Trusted AI, cubes graphic](https://medias.questel.com/pexels_googledeepmind_18069158_271bfa8d7e.jpg)
We need to inform users whenever there is an AI’s use and uncertainty regarding the AI's responses and provide them with the means to verify these responses. This will be achieved by providing sources and explaining the reasons behind the AI's answers.
![Trusted AI, cubes and spheres graphic](https://medias.questel.com/medium_Questel_trust_800217f912.jpg)
When gathering data to train and assess our models, it's crucial to acknowledge the source of the data and confirm that we have permission to use it (such as data that is open-source or provided by users). Additionally, we need to clearly disclose when content has been autonomously generated by an AI.
All our models are fine-tuned and optimized using public information (patents, trademarks and designs texts) or our own data.
When it comes for you to provide information that will be used by our AI inferences, for instance it could be a question, an invention’s description or an action query, information that is often related to clients’ confidential information, we commit to:
- Not using this information to train or fine-tune our models.
- Providing data encryption at rest (AES-256) and in transit (TLS 1.2+)
- Accessing to your business data stored on our systems with a limitation to (1) authorized employees that require access for engineering support, investigating potential platform abuse, and legal compliance and (2) specialized third-party contractors who are bound by confidentiality and security obligations, solely to review for abuse and misuse.
Using this information only with third-party solutions, via APIs, that also comply with all the aforementioned conditions. This is the case for
Hence, we remove the risk of premature self-publication, self-disclosure or revealing information relating to the representation of a client, as our AI inferences have not the possibility to accidentally regurgitate sensitive information provided by yourself.
![Trusted AI, abstract design](https://medias.questel.com/medium_pexels_googledeepmind_17483874_c9823a4c6b.jpg)
You own your inputs and your outputs. We do not claim any ownership on the data generated through our AI inferences.
You control how long your data is retained. We apply a zero data retention once you leave us. Any deleted conversations are removed from our systems within 30 days, unless we are legally required to retain them.
![Trusted AI, glass architecture](https://medias.questel.com/medium_ownership_bbf4d47fdf.jpg)
In some instances, full automation of processes is ideal, but in others, AI should act as a supportive tool to humans, or human judgment is necessary. It's crucial to find the right balance that enhances human abilities and ensures that these solutions are available to everyone.
We ensure this by establishing an AI oversight committee and by regularly organizing workshops and challenges with our clients on the topic. Every conversation starts with “should we do it” instead of “can we do it”.
![Trusted AI, glass shapes](https://medias.questel.com/medium_shubham_dhage_7c_Mqk_Y31_PU_unsplash_138d1aa4e5.jpg)
Do not hesitate to contact us for more information