ThoughtExchange has a number of features that use generative AI or LLMs.
When engaging a new generative AI vendor, ThoughtExchange follows our Secure and Private by Design vendor review process to ensure they meet our requirements. This includes notifying customers of our new vendor while updating our Data Processing Addendum and Subprocessors.
Note: We're making some changes to how we handle consent for our AI features. We're transitioning from using OpenAI consent to a general AI consent, and this includes the Interview BETA consent.
Additionally, we've introduced a new subprocessor, AWS Bedrock, which uses Anthropic AI.
If you previously consented to using Azure, you'll automatically have access to Anthropic AI as part of this update. However, if your organization prefers not to use Anthropic AI, please contact us at privacy@thoughtexchange.com.
In addition to letting customers know we are using a new generative AI vendor, we ask your leader to opt in before generative AI features are available:
ThoughtExchange ensures that our generative AI providers are not allowed to retain or train their AI models on our customers' data through our enterprise agreements and technical controls.
All AI features include visual prompts letting leaders and participants know they are interacting with Generative AI.
ThoughtExchange Features Using Generative AI:
| Feature Name | Feature Description | AI Model |
| AI Summaries | The AI Summary section provides you with an AI-generated summary of the key ideas shared by your participants. This is a great jumping-off point for exploring your results that can give you some ideas about what stood out for participants and what you might want to explore further. Summaries are fully editable (robots don’t always have perfect grammar) in case you want to include them in a report for participants or stakeholders. | Azure OpenAI - GPT4 |
| AI Topics | Topics provides an interactive bar chart depicting AI-generated theme sets (keyword themes, sentiment themes, and (depending on account settings) AI Topics) as well as any custom theme sets you have applied to the Exchange in the Discover Dashboard. Two theme sets can be selected at the same time, allowing you to compare and understand the overlap that exists between them. You can use the icons in the top right-hand corner to toggle between visualizations showing the average star score of the various topics or the number of thoughts contained in each. | Azure OpenAI - GPT4 |
| Article | Article categorizes and summarizes information from thoughts and/or comments shared by participants. | Azure OpenAI - GPT4 |
| Advisor | Advisor will help guide you to find answers to your most important questions about your results. | Azure OpenAI - GPT4 |
| Survey Deep Dives | Dive into your participants’ reasons behind ratings and Likert scales with a qualitative AI interview designed to refine feedback, gather additional context, and clarify responses. | AWS Anthropic - Claude Instant |
| Interview Engagement (Beta/Prototype) | Gather more nuanced insights and transparent feedback through AI-powered 1:1 interviews. Engage in a more meaningful way and uncover the “why” and solutions to problems in a private setting. | Azure OpenAI - GPT4 |
| Room Theme Set Templates | When theme sets are applied to an Exchange, manually entered keywords are checked and then generative AI is used to place remaining thoughts into relevant themes. | Azure OpenAI - GPT4 |
Please see these links to learn more about our Responsible AI & Trust Program and subprocessors.
If you have any questions about AI Features, feel free to reach out to your ThoughtExchange representative or drop us a line at help@thoughtexchange.com