How many entries are we looking at while summarizing? 

Gainsight examines all Timeline entries from the past six months, regardless of the quantity of entries.

What are the best practices on writing timeline entries to get better AI Cheat Sheet results?

Ensure the timeline entries offer full context rather than brief bullet points and abbreviations. The timeline entry must be clearly outlined and detailed to drive conclusion. 

Use Case: A user reviews a timeline to understand the progress and challenges of a project. Each timeline entry must be clearly outlined, detailed, and phrased to convey key insights, enabling a user to easily derive conclusions about the project's status and necessary next steps. Language models function in a similar way by learning from language inputs and generating narrative summaries, effectively mirroring the human process of reading and comprehending detailed notes to form coherent conclusions.

How long does it take to generate the AI Cheat Sheet summary after adding to the layout? 

Once the AI Cheat Sheet is turned on from the Gainsight AI settings page, the data processing based on the timeline entries may take upto a week. However, the actual timeframe may be shorter.

Once the summary is generated, admins that enabled the setting are notified via an email. Gainsight recommends adding the widget to the C360 Layout after this email is received as the data will be ready for the user.

How long it takes to refresh the summary after clicking the Manual refresh?

It takes 2-4 minutes to refresh the data.

Are the sections fixed or can they change based on data? 

The sections (Background, Renewal Discussion, Strategic Priorities, Risks/Issues, Accolades Key Projects, Key Contacts, User Notes) are fixed.

How is User Notes different from other sections? 

The User Notes section is unique compared to other sections as it allows users to edit and add information. This feature is designed to capture any summary points that may have been omitted from the timeline or missed, ensuring a comprehensive and accurate summary with up-to-date information. This notes added here do not impact the content of the cheat sheet. 

Do all key points need to have a time tag associated with them? 

The time tags are not present for key points under the Background section. For other sections, there are edge cases where the time tags are not very accurate, in such cases we avoid providing the tags.

What does different color of time tags signify? 

The different color signifies the recency of the discussion in Timeline from which the key point is coming. We have given three color tags “tan”, “purple” and “blue”. Blue suggests the timeline was logged less than four weeks ago, purple suggests timeline post was logged between one and three months ago, and tan suggests timeline post was logged between four to six months ago.

Why additional details and links to the original post are sometimes not visible in the elaborate section? 

The details in the elaborate section are derived from previous conversations in the timeline. If the key points are brief or lack detailed discussions, the elaborate section may not display any information. Additionally, there are edge cases where GPT fails to identify the discussion, resulting in a lack of details in the elaborate section.
We identify the post that closely relates to the key point and present the top four most relevant posts. However, if none of the posts closely align with the key point, we provide the closest available option, which may not necessarily be the timeline post from which the details or key points were extracted.
IMPORTANT: Clicking on the original post may redirect users to the timeline page without displaying the actual timeline post. This indicates that the post has been deleted and is no longer available in the timeline. Refreshing the page will also remove the same post from the elaborate section.

Can a user provide feedback on any summary points? 

Yes. Users will be able to “Thumbs Up” or “Thumbs Down” summary points to provide feedback. This allows a user to thumbs down a summary point if it’s inaccurate, irrelevant, or duplicative. All users will be able to see the votes on a summary providing more transparency and communication regarding each summary. After clicking dislike, the user can record more detailed feedback through the feedback section or it is recorded but without additional feedback.

Can a user view/edit feedback they provided previously? 

No. A user can’t view or edit the feedback they provided for a point. The feedback is purely for tracking and helping the model to provide more accurate and precise feedback.

Can a user provide feedback for multiple key points simultaneously? 

No. A user can only provide feedback for a single key point at a time.

Is the AI Cheat Sheet regularly updated to reflect the latest information? 

The summary does not auto-refresh. To update the summary, users need to manually click on the Refresh icon at the top. This ensures that all Timeline activities logged after this time are considered, refreshing the AI Cheat Sheet with the latest summary. However, please note that manual refresh can only be performed once within a 24-hour time window. If no timeline posts are added, deleted, or modified/edited, users will see the last refreshed summary.

If I edit an existing Timeline entry and refresh the cheat sheet, will the new edit be included?

Yes, when you edit or delete a post, it is considered as a change. Manually refreshing the summary ensures that the latest changes are reflected in the new summary.

What are the limitations of the AI Cheat Sheet? 

There are four notable limitations:

• In the Background section, company information such as company details, revenue, year founded, and employee size is sourced from the web. However, it's important to note that this information may not always be up-to-date or accurate in certain scenarios.
• There may be situations where the timeline data for a company is available, but a summary has not been generated for that particular company.
• As this summary is powered by generative AI, there are possibilities of hallucinations, missing information, or factual inaccuracies that can be challenging to distinguish them from accurate information..
• The Elaborate feature helps us deep dive and provide more details around key points along with links to up to 4 original Timeline posts from where this information is coming. However, it's essential to note that there might be the cases where these posts are not entirely accurate or may point to unrelated posts 

How can users provide feedback or report issues? 

 In the AI Cheat Sheet, CSMs can click Like or Dislike to provide feedback on specific points. This feedback helps improve the quality of the content.
Note: Clicking Like or Dislike will not remove any point from the summary. 

Is it possible to limit the users who would see the feature? 

The summary widget needs to be added to summary layouts. Admins can control which CSMs have it enabled, and who gets to see it using layout permissions/filters.

Would this feature be made available to all users by default? Or is it something we could test without fully launching to all users? 

As of now, the feature is available to all users who have access to C360, but it can be controlled by admins through layout permissions and filters to test the feature without fully launching it to all users. Admins can choose which CSMs will have access to the feature, and who can see it, by assigning specific layout permissions and filters. 

Can users share the AI Cheat Sheet with external users?

No, users cannot share the AI Cheat Sheet with external users.