Contents:

• Overview
• Use Cases
• Prerequisites
• Enabling the AI Assistant
• Accessing the AI Assistant
• Using the AI Assistant on the Audit Logs Page
• Using the AI Assistant in FlowBuilder
• Conversations
• Considerations: Permissions, Security, and Data Handling
• Troubleshooting and FAQ
• Summary

 

Overview

The LeanData AI Assistant is a natural-language chat experience that helps admins understand and explain LeanData actions and logic. The assistant is currently available in two places:

• The Audit Logs page, where it helps you investigate how records were processed. Instead of opening individual log entries and cross-referencing graph configuration to figure out why a record was processed a certain way, you can ask questions in plain language and receive answers that combine the record's audit log with the relevant graph topology, match logic, and routing context.
• The FlowBuilder canvas, where it helps you understand, navigate, and diagnose your routing graphs. You can ask what a graph does, trace how a record would route, find the node that checks a specific field, or surface where records bottleneck or error. You can also make edits and updates to your graph through the assistant.

The assistant is built on top of LeanData's routing engine, so it has access to your deployments, your match logic, your graph topology, and the full journey of a record across Orchestration and BookIt audit logs. It can explain single log entries, summarize multi-log journeys, execute audit log searches on your behalf, and answer questions about the graph you have open on the canvas.

Use Cases

The LeanData AI Assistant helps admins quickly understand and explain LeanData routing behavior, and helps them stay oriented inside their routing graphs.

Investigating Routing Outcomes

• Explaining a single audit log in plain language to share with a non-technical stakeholder, such as a sales rep asking why a lead was assigned to them.
• Understanding match decisions on Lead-to-Account, Contact-to-Account, or similar match nodes, including which candidate records were considered and why one was selected over the others.
• Exploring the route not taken by asking what would have needed to be different for the record to take a different edge or land on a different owner.
• Summarizing a multi-log journey across Orchestration and BookIt as one narrative, for example a lead that routes through Orchestration and then books a meeting through BookIt for Forms.
• Searching audit logs with natural language when you do not remember the exact filter or field name, for example "show me all logs for leads from Acme that ended in the catch-all queue this week."

Understanding and Maintaining Graphs

• Summarizing what a graph or a specific branch does before making changes or handing the graph off to a teammate.
• Tracing where a specific record would route through the graph.
• Finding which node references a specific field or condition without clicking through every node on the canvas.
• Reviewing a graph for improvement opportunities, such as LeanData capabilities the graph is not using yet and where a new node would best fit.
• Onboarding newer admins or sales support staff who do not yet have deep familiarity with your routing graphs.

 

Prerequisites

Before the LeanData AI Assistant is available, the following must be in place:

• Sync Audit Logs to the cloud. Required to use the assistant on both the Audit Logs page and in FlowBuilder. Your workspace must have audit logs synced to the cloud and the audit log interface enabled. See the Audit Logs Guide for instructions.
• LeanData Remote Site Setting in Salesforce. Your Salesforce org must have the LeanData Remote Site installed so the assistant's endpoints are reachable. For setup steps, see LeanData Admin - Adding the LeanData Remote Site Setting.
• Initially enabling or disabling the assistant requires the LeanData Custom Objects Full Access permission set.
• No separate API key is required. The assistant runs on a key that LeanData provides and manages, on both the Audit Logs page and in FlowBuilder.

 

Enabling the AI Assistant

The LeanData AI Assistant is a self-service toggle under AI Tools. To enable the assistant:

1. From the LeanData app, navigate to Admin > Settings > AI Tools.
2. Toggle AI Assistant (Beta) on.
3. Click Save.

Once enabled, the assistant becomes available on the Audit Logs page to users who have access to the audit logs page, and on the FlowBuilder canvas to users who can open graphs in FlowBuilder.

The AI Assistant (Beta) toggle is a single setting that enables the assistant across the LeanData app. Enabling this turns on the LeanData AI Assistant for both Audit Logs and FlowBuilder.

AI Model and Key

The assistant runs on Google's Gemini model using a key that LeanData provides and manages. This applies to the assistant on both the Audit Logs page and in FlowBuilder. There is no separate API key to provision and no third-party model configuration required.

 

Accessing the AI Assistant

On the Audit Logs Page

To open the assistant:

1. From the LeanData app, click Audit Logs in the left sidebar.
2. In the bottom right corner of the page, click the AI Assistant chat icon.

The assistant opens as a floating chat panel over the Audit Logs page. The panel stays in view as you search, scroll, and open individual log entries, so you can reference logs in the page while you ask the assistant about them.

 

You can also launch the assistant with a log pre-loaded as context. On any audit log entry, click Explain this Log in the Object panel. The assistant opens with that log already attached to the conversation, ready to answer questions about it.

In FlowBuilder

To open the assistant from within FlowBuilder:

1. From the LeanData app, open a routing graph in FlowBuilder.
2. Click the AI Assistant chat icon on the canvas.

The assistant opens as a floating chat panel over the canvas. You can move and resize the panel so the nodes you are asking about stay visible while you chat.

Using the AI Assistant on the Audit Logs Page

You interact with the assistant by typing questions into the chat panel. The assistant uses your audit log data, your active graph deployments, and LeanData's routing logic documentation to generate answers. Answers include deep links to the relevant log entries, graphs, and Salesforce records so you can verify the explanation against the underlying data.

Asking About a Specific Log

With a log loaded into the conversation, you can ask questions such as:

• What happened to this record?
• Why was this lead routed to this owner?
• Why was this account matched instead of one of the other candidates?
• Which fields on this record drove the routing outcome?
• What would have needed to change for this record to be routed to a different owner?

The assistant responds with a plain-English explanation that references the specific nodes, edges, and match decisions in the graph that processed the record. Where helpful, the response links directly to the relevant node in the graph, the audit log entry, or the Salesforce record.

Searching Audit Logs with Natural Language

You can ask the assistant to find logs without using the structured search filters. For example:

• Show me all logs for leads from Acme this week that ended in the catch-all queue.
• Find logs where the Round Robin node skipped the assigned rep in the last 24 hours.
• List all audit logs related to johnsmith@companyname.com in both Orchestration and BookIt.

The assistant translates the request into a structured search, runs it against your cloud audit logs, and displays the matching results inside the Audit Logs page. You can refine the search by following up in the same conversation, for example "narrow this to the last 7 days" or "only show me leads, not contacts."

Summarizing a Record's Journey

When a record has been touched by multiple LeanData processes, you can ask the assistant for a single narrative summary that stitches the logs together. For example:

• Summarize the full journey of this lead across all LeanData processes.
• What happened to this lead from the time it was created to the time it booked a meeting?

The assistant produces a chronological narrative across Orchestration and BookIt logs, with links to each underlying log entry.

Drafting an Explanation for a Non-Technical Stakeholder

The assistant can rephrase its explanation at different levels of technicality. After getting an answer, you can ask:

• Rewrite this in a way I can send to the sales rep.
• Make this shorter and remove the graph terminology.

The assistant returns a version of the explanation suitable for forwarding to a non-admin audience.

Suggested Prompts

Suggested prompt buttons appear in the chat panel and on individual log entries to launch common questions with a single click. Quick actions include prompts such as Explain this log, Why this owner?, and Summarize this journey. Clicking a quick action sends the prompt to the assistant with the relevant log already attached as context.

 

Using the AI Assistant in FlowBuilder

On the canvas, the assistant is grounded in the graph you have open. It combines the graph's topology, node configuration, and execution data with LeanData's routing logic documentation to answer questions about that specific graph. Where helpful, the assistant highlights the relevant nodes directly on the canvas so you can click into them. While the assistant works on a question, the panel shows its progress and the steps it is taking, so you can follow how it arrives at an answer.

The panel keeps conversation context within a session, so follow-up questions work naturally. For example, after asking for a graph summary, you can ask "now explain the branch on the left" without restating anything.

Understanding a Graph

Ask the assistant to explain the graph or a part of it:

• What does this graph do?
• Summarize this graph so I can hand it off to a teammate.
• Explain the branch that handles enterprise leads.

The assistant responds with a structured summary in chat, and the nodes it references will be highlighted on the canvas so you can see which part of the graph each point describes.

Tracing a Record's Path

Ask where a specific record would go:

• Where would this lead go?
• Which owner would a contact from Acme end up with?

The assistant traces the path through the graph and highlights it on the canvas, so you can follow the record's route edge by edge.

Diagnosing Bottlenecks and Errors

Ask diagnostic questions about how records are moving through the graph:

• Where are records getting stopped in this graph?
• Where are failures concentrated?
• Which paths have no traffic?

The assistant returns a ranked list of nodes by error or traffic volume. Each result is clickable, so you can focus the canvas on the node and investigate.

Finding Nodes and Configuration

Ask the assistant to locate logic inside the graph:

• Which node checks Industry?
• Where do we filter out records with personal email domains?

Matching nodes will be highlighted on the canvas, and you can click one to select it and open its configuration.

Learning How to Build

Ask the assistant how to accomplish something in FlowBuilder:

• What node should I use to distribute to my EMEA team?
• Which node should I use to hold records for a working-hours window?

The assistant walks you through the approach and points you to relevant capabilities, such as an existing Snippet that already implements the pattern.

Getting Recommendations for Your Graph

Ask the assistant to review the graph for improvements and suggestions:

• What LeanData capabilities am I not using in this graph today?
• Where can we improve how leads are routed?

The assistant reviews the graph and suggests capabilities that fit your routing, such as lead to opportunity matching, account hierarchies, skills-based round robin, or an AI Inference Node for use cases like lead tiering or sentiment analysis on form comments. When a suggestion involves adding a node, the assistant recommends where in the graph to place it, and you can ask it to bring that node into the graph as a proposed edit for your review.

Suggested Prompts

Predefined prompt buttons appear at the top of the chat panel to launch common questions with a single click, including prompts such as Summarize this graph, Explain this path, and Where are records not flowing?

 

Conversations

The assistant organizes your interactions as conversations. Each conversation keeps the assistant's context from one question to the next, so follow-up questions reference the same logs, the same graph, and the same line of inquiry.

• Start a new conversation. Click New Conversation in the assistant panel to clear the context and begin a fresh investigation.
• Switch conversations. Click the conversation list icon in the assistant panel to view your saved conversations and reopen one.
• Rename a conversation. Click the conversation title to give it a more descriptive name, such as "Acme catch-all investigation."
• Delete a conversation. Open a conversation, click the overflow menu, and choose Delete. Deleted conversations are removed from your history.

Conversations are scoped to your user. Other admins in the org do not see your conversation history. A conversation you start on the Audit Logs page can be picked up from the assistant panel in FlowBuilder, so you can carry an investigation from your logs onto the canvas without restating the context.

Your conversation and context are maintained as you navigate between tabs and sections of the LeanData app. Closing the assistant or logging out of Salesforce ends the active session, and the conversation remains available in your saved history.

 

Considerations: Permissions, Security, and Data Handling

• Access. Any user who can access the Audit Logs page can use the assistant there, and any user who can open a graph in FlowBuilder can use the assistant on the canvas, once it is enabled for the org. Only users with admin permissions to LeanData settings can enable or disable the assistant.
• Permission-aware responses. The assistant respects your Salesforce permissions and LeanData role-based access. It only reads data you can already access and never elevates your permissions. For example, if you can see a log entry but do not have access to the underlying lead record, the assistant shows the log data but does not link into or analyze that record.
• Data sent to the model. When you ask a question on the Audit Logs page, the relevant audit log records, the relevant graph configuration, and LeanData's routing documentation are sent to the model that powers the assistant. When you ask a question in FlowBuilder, the configuration and execution data of the graph you have open are sent instead. The model generates a response and returns it to the assistant.
• Data residency. The audit log data the assistant references is the same data already synced to LeanData's cloud audit log storage. The assistant does not introduce additional copies of your CRM data outside of what Audit Logs already stores.
• Model usage. The assistant runs on the LeanData-provided key on both the Audit Logs page and in FlowBuilder. Any data sent to the model is not used to train the model and is not shared with others.
• Rate limits. To protect the service and manage cost, the assistant applies per-user and per-org rate limits. If you reach a limit, the assistant pauses new requests until the limit resets and surfaces a message in the chat.
• Personally Identifiable Information (PII) handling. Audit logs can contain PII such as names, email addresses, and company information. This data is sent to the model only as needed to answer your question and is handled according to LeanData's data processing terms.

 

Troubleshooting and FAQ

The AI Assistant icon is not visible on the Audit Logs page.

• Confirm that Audit Logs have been synced to the cloud. The assistant does not appear on the legacy per-product audit log pages.
• Confirm that an admin has toggled AI Assistant on under Admin > Settings > AI Tools.
• Confirm that your user has access to the Audit Logs page.

The AI Assistant icon is not visible in FlowBuilder.

• Confirm that an admin has toggled AI Assistant on under Admin > Settings > AI Tools.
• Confirm that your user can open graphs in FlowBuilder.
• Confirm that Audit Logs have been synced to the cloud. Cloud audit log sync is required for the assistant in FlowBuilder, the same as on the Audit Logs page.

The assistant says it cannot find a log I expect it to find.

• Confirm the date range. Cloud audit logs are retained for the configured retention period.
• Try referencing the record by a more specific identifier, such as the Salesforce ID, rather than a name or partial email.

The assistant gives an explanation that does not match what I see in the graph.

• Confirm that the graph deployment shown in the audit log entry matches the deployment you are reviewing. The assistant explains the log against the deployment that processed the record, which may not be the current active version of the graph.
• If the explanation still appears incorrect, use the thumbs-down feedback control in the chat panel to flag the response. Feedback is reviewed to improve the assistant.

Can I bring my own model API key?

• No. The assistant currently runs on the LeanData-provided key on both the Audit Logs page and in FlowBuilder.

Can the assistant make changes to my graph?

• Yes. In FlowBuilder, the assistant can help create and edit nodes. Review AI-suggested changes before saving or deploying graph updates.

Is my data used to train the model?

• No. Data sent to the model is used only to answer your question. It is not retained by the model provider for training and is not shared. Please use the thumbs-up / thumbs-down button by the assistant's responses if you would like to provide feedback to improve future iterations of the assistant.

 

Summary

The LeanData AI Assistant gives admins a natural-language way to investigate routing behavior and understand their graphs. On the Audit Logs page, it combines your cloud audit log data with your graph deployments and LeanData's routing logic to answer questions in plain language about why a record was processed a certain way, summarize multi-log journeys, and execute audit log searches on your behalf. In FlowBuilder, it answers questions about the graph you have open, traces record paths, finds nodes and configuration, and surfaces where records bottleneck or error.

For questions or additional assistance, please contact LeanData Support.