AI-Powered Discovery in Process HQ

Overview

Process HQ users can gain new insights into their data by asking questions to Appian AI Copilot. Users get faster, real-time insights into their data without relying on a human being to find and analyze the data for them.

These capabilities are only available in select regions. If your Appian environment is in an unsupported region, you can choose to send your data to a supported region. This won't change your environment's region. For more information, contact Appian Support.

AI Copilot for data fabric

In the Data Catalog or Process HQ home page, users can ask questions to understand their data, and partner with AI Copilot to generate new reports from datasets.

Analyzing data

Users can also ask AI Copilot to filter or analyze data from any dataset. For example:

• Filter: What cases have been created this year?
• Analyze: How many of those cases are open?

Users don't need to know which dataset contains the information they're looking for in order to get started. Instead, they just click AI COPILOT and ask their question in their own words. AI Copilot queries the appropriate dataset to provide insights from the available information. The response also includes information about how AI Copilot arrived at its answer, ensuring trust and accuracy.

Scroll or drag to pan · ⌘+/⌘− to zoom · Esc to close

AI Copilot can interpret questions and provide answers in many languages. As with all generative AI, you may experience a range of responses depending on factors like language, phrasing, and complexity of the request.

If you ask a question that references a field value, we recommend wrapping the value in quotation marks so AI Copilot is able to reference it. For example, if your record data is in English, you would format a question in Spanish as: ¿Cuantos casos están "Open"?.

Generating reports

AI Copilot can generate a report based on a dataset. Users can ask for any report they'd like to see, and AI Copilot will generate a preview of that report right in the chat window.

Scroll or drag to pan · ⌘+/⌘− to zoom · Esc to close

Users can continue chatting to make changes to the report. Then, they can click CREATE REPORT to make any last adjustments manually, save, and share the report.

AI Copilot for reports

Users can ask AI Copilot questions about their reports, or select from a list of suggested questions to discover potential areas of improvements and next steps. AI Copilot is available for reports that involve grouped data or basic aggregations, such as sum, count, or average. AI Copilot for reports cannot perform calculations and is not intended for scenarios that require numerical accuracy. If you want to ask AI Copilot questions about your data that require calculations (for example, "What cases have been created this year?"), use AI Copilot for data fabric instead.

Scroll or drag to pan · ⌘+/⌘− to zoom · Esc to close

Ad hoc event detection

Ad hoc events, like "John Doe commented on an order" or "Jane Smith updated an employee," are typically not useful when analyzing how the process itself is doing. As a result, ad hoc events should be excluded from event data that will be used in process insights.

In process insights, AI Copilot can automatically detect ad hoc events and suggest filters to help you remove them. If any potential ad hoc events are detected, a prompt will appear to help you filter out unwanted events.

Suggested views

An important step towards exploring your enterprise data in Process HQ is to create a view. But knowing which attributes, details, and filters to choose can be intimidating and time-consuming.

Using AI Copilot, you can create views more easily, spending less time configuring your process and more time exploring your data. AI Copilot uses what it knows about your process data to provide multiple suggestions for new views. As you explore each suggested view, you're able to edit the suggested attributes, process details, and filters.

Suggested KPIs

System-provided KPIs in Process HQ cover most common use cases for measuring process success and tracking relevant indicators. However, in addition to these, AI Copilot can suggest KPIs based on a high-level evaluation of your specific data.

When you select a suggested KPI, the KPI definition and filters are automatically configured, then provided for your review and editing as necessary. With system-provided KPIs and AI-generated KPIs, you'll quickly and easily find KPIs to address all uses cases for your business.

Generated insight summaries

As you drill down on the characteristics of your process, you combine the analyses you see into an insight—that is, an actionable conclusion you're ready to start sharing with your colleagues. To speed up communication about that insight and facilitate collaboration across your organization, AI Copilot generates a short text summarizing the most important aspects of the insight.

You can review and edit this summary, so you can be confident that the summary is the best ambassador for the insight.

Troubleshoot AI Copilot

If AI Copilot isn't available, ensure your environment meets the security and compliance requirements and that these features are enabled in the Admin Console.

If AI Copilot doesn't appear in the data catalog, it may be because:

• Your environment doesn't have the appropriate license. Contact your administrator.
• More than 250 records types are configured to display as datasets in the data catalog. Configure between 1 and 250 records types to display in the data catalog.

If AI Copilot cannot answer a question in the data catalog, it may be because:

• The service that powers AI Copilot is down or experiencing high usage. Users should wait and try again later.
• Questions are too vague. Users should be more specific about their questions so AI Copilot can query the appropriate datasets.
• AI Copilot doesn't recognize usernames in the configured format. Contact Appian Support to request an update to your site's properties to use a different format.
• The user has exceeded the token limit. This will require the user to start a new conversation.

If AI Copilot cannot analyze a report, it may be because:

• The report does not contain any aggregated data. Users must either group data or apply a calculation (like sum, count, or average) to aggregate data for analysis.
• The report is too large. Users should remove fields or apply a filter to reduce the number of rows.
• The user has exceeded the token limit. This will require the user to start a new conversation.

Sessions and token limits

A session is a conversation between a user and AI Copilot. During a session, a user can ask AI Copilot questions about their data. At some point, a user may receive a message from AI Copilot indicating that the conversation has reached its limit. This occurs when the user exceeds the context window for a given session.

If a user hits the limit, they will need to start a new session to continue asking questions. Users will either be prompted to start a new session, or they can click Restart your conversation with AI Copilot in their report or Clear conversation in the data catalog to manually start a new session.

AI Copilot won't remember questions from a previous session. If a user starts a new session, AI Copilot won't have any context about the previous conversation. This means AI Copilot won't have answers to follow-up questions from a previous session. Users will need to provide AI Copilot with the necessary context at the beginning of each session.

Learn more about AI Copilot.

Feedback