Tonkean AI FAQs
This topic provides answers to frequently asked questions about Tonkean AI, including how Tonkean uses AI, how data is used and stored, and how AI features work. This topic also provides links to relevant sections where admins can configure these AI features and ensure your workflows get the most out of Tonkean AI power.
How does Tonkean leverage artificial intelligence (AI)?
AI is integral to the Tonkean platform—the most obvious features being the AI Front Door and the AI Concierge, where internal users can use plain language to ask questions, submit requests, and get context-sensitive help. These features combine to form the Enterprise Copilot, a suite of generative AI capabilities that streamline and support the process experience for requesters and other stakeholders.
Tonkean generative AI features use customer data as reference and context, not as training material.
Tonkean has numerous other AI capabilities that work behind the scenes in the module workflow:
Training sets are a natural language processing (NLP) feature that allows you to extract and classify text from documents and use that text in your modules.
Optical character recognition (OCR) is an important machine learning component of many AI-powered workflows, allowing you to identify and extract text from a digital image like a PDF.
Native integrations with popular AI providers like OpenAI, Google, and Anthropic offer access to a wide range of generative AI and machine learning models (LLMs), including GPT-4.
What is the AI Front Door and how does it work?
The AI Front Door is the go-to place for internal users to ask questions and submit requests from your shared services teams, without needing to understand complex processes or policies. The AI Front Door is powered by Microsoft Azure OpenAI. It leverages the forms and data sources in your solutions as context to answer questions, initiate processes, and perform other tasks configured in your modules.
The AI Front Door does not train the model on your data. Instead, it selectively and strategically accesses the data in your enabled forms and data sources to act as context and reference to serve requesters. For example, if a member of the sales team asks how long the opportunity with ACME Inc. has been open, the AI Front Door locates the record for ACME Inc. in your connected Salesforce data source, calculates the number of days it's been open, and returns that to the user. For a more detailed explanation of the different ways the AI Front Door searches for information in your Tonkean instance and connected systems, see Set Up the AI Front Door.
To set up and begin using the AI Front Door in your process, contact Tonkean Support or your dedicated customer success manager.
The process can vary depending on the request type, but generally goes through the following steps for a basic question:
A user submits a question or request: "Do we have an NDA for Corellia Labs?"
Tonkean identifies keywords and phrases ("NDA" and "Corellia Labs") and then sends them to the LLM (either the default model or your preferred connected model).
The LLM analyzes the keywords and phrases in the context of your connected data sources, interfaces, forms, and other content and finds the content or information that has the highest probability of being a match for the request. It may find, for example, the existing mNDA with Corellia Labs stored in your OneDrive instance.
The LLM generates a plain-language response with a link to the NDA if it exists or, if there's a process for creating a new NDA, provides the option to begin that process.
Tonkean replies to the user with the LLM-generated information.
See the diagram below for a simplified version of the process:
How can I enable the AI Front Door to access my content?
By default, access via the AI Front Door is turned off. To enable access, you must navigate to each resource and enable Tonkean AI to access it. For more information about enabling access, see Enable AI Front Door Access to Your Content.
Once access is enabled for the AI Front Door, anyone with access to the AI Front Door has access to the data contained in that data source, regardless of the permissions in the third-party application or system.
How can I improve the accuracy of the AI Front Door?
The accuracy of the AI Front Door is directly related to the amount of information in the AI-enabled interfaces, forms, and data sources in your solutions. The more detailed and complete the information, the more accurate the Front Door will be.
There are specific steps you can take to improve Tonkean AI accuracy and results for different kinds of resources (interfaces, forms, data sources, etc.). For more information about improving the accuracy of the AI Front Door, see Optimize Content for the AI Front Door.
Does Tonkean use my data to train its models?
No, Tonkean does not use proprietary models—nor does it use customer data to train any AI models. Tonkean AI users your connected data sources, interfaces, and other resources as context and reference to help users.
By default, Tonkean AI features use Microsoft's Azure OpenAI Service, which includes pre-trained models like GPT-4. Customers have the option to use an alternate LLM or generative AI model to power Tonkean AI features.
Do I retain ownership of the data used in Tonkean AI features?
Yes, you always retain ownership of your data. We do not use or retain customer data used to facilitate AI features like the AI Front Door and AI Concierge for any purpose other than providing these services.
Your usage of connected generative AI service providers like OpenAI or Google's Gemini may be subject to their specific terms and conditions.