We’ve migrated our documentation to a new site, which means some URLs have changed.
Subscriptions
Subscriptions
Breadcrumbs

Tutor

How Tutor Works

When you ask a question or make a request to Tutor, it performs the following steps:

  1. Product Detection: Tutor first determines if your question pertains to a specific Piano product and filters relevant documentation accordingly.

  2. Semantic Similarity Search: It assesses the semantic similarity of your question to available documentation and Support FAQs.

  3. Source Selection: Tutor selects the top 15 most relevant sources from its knowledge base.

  4. Response Generation: Using LLMs, Tutor constructs a coherent answer based on the selected sources.

  5. Source Listing: Tutor provides a list of the sources used to generate the answer, allowing you to verify the information. The Support FAQs also indicate the age of the content to help you assess if the information might be outdated.

Getting Started with Tutor

Accessing Tutor

Piano clients can access Tutor at tutor.piano.io. All individuals with logins to Piano Piano Subscriptions or Piano Analytics production applications should be able to log in to Tutor using their existing credentials.  If a user at a client organization does not have a login, they should be added to the production application, and within 24 hours that user should be able to log in to Tutor.

Clients using the legacy Piano Analytics Single Sign-On (SSO) may also have login issues. The primary solution to this issue is to migrate to the newer SSO solution, but such users should be able to log in using the "reset password" link. Currently, Piano clients with only a SocialFlow account cannot yet log in to Tutor. Work is planned to unify the SocialFlow SSO to make this possible in the future.

If you experience any issues accessing Tutor, please contact your account manager or reach out to tutor@piano.io.

Using Tutor Effectively

Asking Questions

  • Be Specific: Provide clear and specific questions to receive the most accurate answers. Avoid ambiguity and unnecessary details.

  • Provide Context: Include relevant background information and details about the product or feature you are inquiring about.

  • Mention Piano Products: Explicitly mention the Piano product (such as Management + Billing, Composer, ID, Piano Analytics, DMP, ESP, or SocialFlow) to filter the relevant information accordingly.

  • Use Keywords: Use relevant keywords related to the topic, such as product names or features.

  • Avoid Links: Instead of including links, copy and paste the relevant text snippet you want to refer to.

  • Break Down Questions: If you have multiple questions, break them down into separate requests.

Example Questions

Questions that are specific and include relevant names and keywords are likely to produce the best results:

  • "Which Management + Billing API can I use to fetch subscriber details?"

  • “In Identity Management, how long does a password reset link last for a user?"

  • "How do the different source properties in Piano Analytics differ?"

Ambiguous questions like "Tell me about reports," which could relate to various products across the Piano platform, are likely to produce muddled answers. Similarly, using single keywords like "API," "login," or "conversions" may pull up documentation from various Piano products.

For better results, try more specific questions like:

  • "How can I see conversions for a specific Composer experience?"

  • "What metrics in Piano Analytics are commonly used to track conversions?"

Citations

After answering a question, Tutor will automatically generate a "How this answer was produced" message that explains which sources were used to arrive at the answer. This feature is intended to make it easy to verify Tutor's response.

Please note that Tutor can sometimes provide inaccurate information due to outdated documentation, ambiguous content, or limitations of the language model. If you receive an inaccurate answer, you can follow up by emailing tutor@piano.io so we can address any gaps in the documentation and continue to improve Tutor's accuracy.

Rating Answers

After receiving an answer from Tutor, please rate its usefulness:

  • Very Useful: Tutor was accurate and provided a correct answer.

  • Somewhat Useful: Tutor was partially correct or slightly off the mark.

  • Not Useful: Tutor was unable to answer or provided an incorrect answer.

You can also provide more detailed feedback by emailing the Tutor team at tutor@piano.io.

Follow-Up Questions

You can ask follow-up questions to delve deeper into a topic or clarify your initial request. Tutor will use your previous questions as context to refine its answers.

After you ask a question, you will see a "Suggest related questions" option below your query. Clicking this button will prompt Tutor to suggest 5–10 alternative questions related to your original one.

Restarting the Chat

If you want to start a new topic, use the "Clear chat" option to reset the context.

History

The "History" tab allows you to view your past Tutor conversations if you want to revisit an old answer. By clicking on an individual thread, you can share the URL in your browser with other users within your organization. This allows colleagues to access the conversation and benefit from the information shared. They will not automatically see the questions you ask in their own history tab, but they will be able to access your Tutor thread if you share the direct link with them.

Please note:

  • Sharing Within Your Organization: Only users within your organization with access to Tutor can access the shared links. This feature facilitates collaboration by allowing team members to share knowledge and solutions easily.

  • External Access: Users outside of your organization will not be able to access these shared links. This ensures that your organization's data and conversations remain secure.

If you need Piano's support team to double-check a Tutor answer, you can share the URL of the Tutor thread in your email to support@piano.io. This will provide support with the relevant context for your request.

Multi-Lingual Use

Thanks to the capabilities of LLMs, Tutor can converse in your language of choice. However, not all parts of the Tutor interface are localized, so while the question and answer may be in any language, citations and the sources themselves will remain in English.

Contacting Support

If you need Piano's support team to double-check a Tutor answer, please include the URL of the Tutor thread in your email to support@piano.io. You can find the URL of the Tutor thread by going to your "History" tab, clicking on the relevant thread, and copying the URL from your browser. This will provide support with the relevant context for your request.

You can also contact the Tutor product team by emailing tutor@piano.io to provide feedback or submit Tutor-related feature requests.

Transparency Notice

Tutor is an AI-powered chatbot. This notice is provided to ensure transparency and inform you about the use of artificial intelligence in this service.

Nature of the System

Tutor is an AI system designed to assist users by providing automated responses to questions and requests related to Piano products. The system is based on natural language processing and machine learning technologies. It is a third-party supported system powered by Anthropic and OpenAI LLMs (Large Language Models). We query these models via API (Application Programming Interface) by providing:

  • Client questions

  • Relevant Piano-generated documentation to help the models understand the context

  • Sanitized data with personal data removed and substituted with placeholder values for email-based support requests

It is a limited-risk AI system under the EU AI Act, meaning it poses minimal potential risk to users' safety or fundamental rights.

Purpose and Capabilities

Tutor is used to answer frequently asked questions and to provide information about our services and products. It can be used to assist with basic troubleshooting or navigation.

With this tool, we're aiming to provide efficient support with a smooth user experience while still allowing any users who want to be routed to our human support team to do so.

Tutor does not make legally binding decisions or replace professional human advice. For complex or sensitive issues, a human representative will be available.

AI Identification and Human Oversight

Tutor is not fully comprehensive, in the same vein as other types of AI models – it can only generate answers based on content and is not capable of true problem-solving or applying human logic, which means the answers generated are only as good as the information it has access to. This interaction is automated and generated by an AI system. If you prefer to speak with a human, you can request human assistance at any time by emailing support@piano.io.

Tutor is monitored by Piano to ensure the accuracy and appropriateness of its responses. Conversations may be reviewed to improve system quality and ensure compliance.

Processing of Personal Data

When generating answers to support requests sent via email, Tutor takes steps to remove all personal data (such as names, emails, user IDs, and other personally identified information) within its EU data center before using external APIs.

Piano’s agreements with LLM vendors stipulate that data sent via API will not be used for the training of these vendors’ underlying models. Piano tracks and analyzes its clients’ use of Tutor to improve the product, provide more relevant answers, and enhance the experience of Piano customers.

User Rights

As a user interacting with this AI system, you have the right to:

  • Be informed you are interacting with an AI (this notice)

  • Refuse further interaction and request human contact

Contact

If you have questions or concerns about this AI system, please contact support@piano.io.

Last updated: