Beginning with release 18.1, Issuetrak offers an AI Note Assistant. When enabled, this feature makes an "AI Assistant" dropdown available in the note editor to choose from several possible transformations of any note that you are adding or editing.
When an agent clicks on AI Assistant, several options appear:
- Fix grammar & spelling
- Make more professional
- Make friendlier
- Summarize
- Expand & clarify
When you click one of these options, your configured LLM will transform the entire note in the note editor. It will base the changes on the option you selected, the existing note content, and the prompts that are configured for the LLM in the AI Note Assistant settings.
This feature requires that you have AI enabled and configured in your site.
Activating AI Note Assistant
You can enable the AI Note Assistant feature by doing the following:
Steps:
- Click the gear icon in the top right to open the Settings lightbox, then click on AI Settings.
- Click on the Note Assistant tab.
- Toggle Note Assistant.
- Click Save.
The "AI Assistant" dropdown will now appear in the note field for your agents when they add or edit a note. It is ready to be used.
Prompt Configuration and Tweaking
While not strictly necessary, further configuration can help you fine-tune how the resulting notes are changed by the Note Assistant. You can heavily influence the formatting, tone, verbosity, and even the persona that each overall note takes on.
What does the configuration look like, though?
Steps:
- Click the gear icon in the top right to open the Settings lightbox, then click on AI Settings.
- Click on the Note Assistant tab.
- The first 2 settings on the Note Assistant tab have explanations on-screen, which we will duplicate here:
- Max Context Tokens - Maximum issue text sent to the AI at once (a token is roughly 3/4 of a word). Notes over the limit are rejected rather than truncated, in order to avoid mangling the original content.
- Max Output Tokens - Maximum length of the rewritten note. Keep this at least as large as Max Context Tokens - a rewrite is usually about the size of the original note, and "Expand & Clarify" makes it longer. A rewrite that hits this limit is cut off mid-note.
- Expand and inspect each of the prompt textboxes.
- You can cancel your changes to any of the prompts by clicking on the X icon above its respective textbox.
- You can revert a prompt to the "stock" verbiage by clicking the back arrow icon above its respective textbox.
- You can choose to edit any of the prompts by clicking the pencil icon above its respective textbox.
There are 7 prompts that you can adjust for this feature:
- Plain Text Header Prompt
- Plain Text Footer Prompt
- Fix Grammar & Spelling Option Prompt
- Make More Professional Option Prompt
- Make Friendlier Option Prompt
- Summarize Option Prompt
- Expand & Clarify Option Prompt
In the next section, we will go over what each of the prompts do to influence the output of the Note Assistant.
Plain Text Header Prompt
This provides the opening instructions sent to the LLM. It defines the assistant's role and the rules for handling the note content, and is combined with the selected option's prompt and the Footer Prompt (next section) to form the full instructions for every rewrite.
You can use {issue} or {issues} anywhere in your text, and they will be automatically replaced by your custom labels (e.g., "Ticket").
Plain Text Footer Prompt
This closes the instructions sent to the LLM. It defines the rules for the rewritten note: the output format, keeping facts and identifiers unchanged, writing in the same language as the original note, and returning the note unchanged when there is nothing to transform. The LLM's entire response is inserted directly into the note field, so these rules are what keeps the output safe to use as-is.
You can use {issue} or {issues} anywhere in your text, and they will be automatically replaced by your custom labels (e.g., "Ticket").
Fix Grammar & Spelling Option Prompt
This describes the task the LLM performs when the user picks "Fix Grammar & Spelling". It is inserted between the Header Prompt and the Footer Prompt, so it should state only what this option changes about the note and what it leaves alone.You can use {issue} or {issues} anywhere in your text, and they will be automatically replaced by your custom labels (e.g., "Ticket").
Make More Professional Option Prompt
This describes the task the LLM performs when the user picks "Make More Professional". It is inserted between the Header Prompt and the Footer Prompt, so it should state only what this option changes about the note and what it leaves alone.
You can use {issue} or {issues} anywhere in your text, and they will be automatically replaced by your custom labels (e.g., "Ticket").
Make Friendlier Option Prompt
This describes the task the LLM performs when the user picks "Make Friendlier". It is inserted between the Header Prompt and the Footer Prompt, so it should state only what this option changes about the note and what it leaves alone.
You can use {issue} or {issues} anywhere in your text, and they will be automatically replaced by your custom labels (e.g., "Ticket").
Summarize Option Prompt
This describes the task the LLM performs when the user picks "Summarize". It is inserted between the Header Prompt and the Footer Prompt, so it should state only what this option changes about the note and what it leaves alone.
You can use {issue} or {issues} anywhere in your text, and they will be automatically replaced by your custom labels (e.g., "Ticket").
Expand & Clarify Option Prompt
This describes the task the LLM performs when the user picks "Expand & Clarify". It is inserted between the Header Prompt and the Footer Prompt, so it should state only what this option changes about the note and what it leaves alone.
You can use {issue} or {issues} anywhere in your text, and they will be automatically replaced by your custom labels (e.g., "Ticket").
