Frequently Asked Questions
How does Contract Express Chat work?
Contract Express Chat uses OpenAI’s GPT-3.5 technology paired with a RAG (Retrieval Augmented Generation) mechanism to access data from our own Contract Express documentation. When you pose a question to the chat, it will retrieve the most relevant data from the database and generate a personalized response using GPT-3.5, ensuring that the information is relevant and tailored to your query.
The database includes information from the Contract Express Author Help manual, the Knowledge Base, the Author and Admin manuals as well as some quick guides.
How can I get access to Contract Express Chat?
Contract Express Chat will be available to all Contract Express cloud customers on v9.15 and above. Make sure you have the latest version of the Author tool installed, and that you are pointing to your production server on the latest version.
I am hosted on-premises, can I get access to Contract Express Chat?
Contract Express is not available through on-premise production environments. Customers who want to experiment with Contract Express Chat can create a trial account on one of our sandbox environments. Please contact support for details at
contractexpress-support@thomsonreuters.com.
What Azure OpenAI services are used by Contract Express Chat?
Where possible, each Contract Express instance is routed to an Azure Open AI service in the region where it is hosted. If the necessary Azure OpenAI service is not available in the same region, it is routed to the nearest available Azure OpenAI service (e.g., AI requests from instances hosted in the Germany region are routed for processing to France region). This process does not involve data storage.
How can I get the most out of Contract Express Chat?
See below for some quick tips to help you maximise your experience with Contract Express Chat:
Ensure that you are clear and concise in your queries, providing relevant detail and context where possible.
Questions do not need to be written in technical language, but if you know certain terminology, that will help narrow down the search for relevant material.
Previous questions posed in the same chat are included as additional context to help guide a response.
If you wish to explore a new topic, start a new chat to reduce the context of your previous queries and maximise accuracy in the responses.
Make use of the feedback options to help us refine our service. Click thumbs up or down on each response to help us improve and enhance the Chat's capabilities and fill out
this survey to give us more details on your experience.
How do I know the answers are correct?
The chat will provide answers that reflect the database linked to the server that you are connected to in the Author tool (see Options > Servers). That means that if you are connected to a server on v9.15, the information will be up to date with information up to v9.15.
Validity is important to us, which is why we included the sources that were used for each answer provided. While we believe the chat is extremely useful, we can’t guarantee it’s not missing anything. It will still be important for you to test your templates to ensure you are building the experience you want. If you do notice something that is wrong or missing, please let us know by filling out this quick
survey.
Will OpenAI be able to use my data?
No, Thomson Reuters’s use of Generative AI is via dedicated API access through trusted partners such as Microsoft Azure and Amazon Web Services. Thomson Reuters expressly prohibits any vendor from retaining or using customer data to train their Generative AI models.
Will my questions be reviewed for product improvement?
Yes, the Contract Express product team will have access to the question prompts, responses and feedback. This information will be stored in the same database corresponding to the Contract Express instance you are connected to. We will review the questions, responses and feedback to improve the Chat’s capabilities.