AI Insights and Interaction History
CX uses AI to surface relationship context that would otherwise require digging through email threads, task logs, and project histories. On every company record, you'll find AI-generated executive summaries, sentiment analysis, and a consolidated interaction timeline. These tools help your team quickly understand where a client relationship stands — especially useful when picking up an account you didn't manage previously, preparing for a client meeting, or investigating a declining Pulse score.
Prerequisites
AI insights require email data to generate. If your team hasn't connected email in CX, or if a company has no recent email activity, AI summaries will not be available for that record.
You need at least clients (read) permission to view AI insights on company records.
AI Executive Summary
The AI executive summary appears on the Overview tab of a company record. It provides a concise, automatically generated narrative of the current state of the client relationship, including:
Recent activity highlights — key emails, completed tasks, and project milestones from the past 30 days
Open items — outstanding tasks, pending requests, and overdue deliverables
Relationship context — tone and frequency of recent communication, any notable changes in engagement patterns
The summary updates automatically as new interactions occur. It is not editable — it reflects what CX has observed in the data.
Tip: The AI summary is particularly useful for managers reviewing accounts they don't directly manage, or for team members covering for a colleague. It gives you a 30-second picture of where the relationship stands without reading through individual email threads.
Sentiment Analysis
CX analyzes the tone of email interactions with each company and provides a sentiment indicator that helps you gauge client satisfaction over time. Sentiment is displayed as a trend on the company record and factors into the overall Pulse health score.
Sentiment analysis considers:
Email tone — language patterns in client emails that suggest satisfaction, frustration, urgency, or neutral communication
Response patterns — how quickly the client responds, whether response frequency has changed, and whether conversations are closing or escalating
Topic patterns — whether recent conversations are routine (census questions, file requests) or escalatory (complaints, deadline concerns, service issues)
Sentiment analysis is directional, not definitive. It's a signal to investigate, not a diagnosis. If sentiment drops, review the recent email history on the Email tab to understand what's driving the change before taking action.
Interaction History Timeline
The interaction history consolidates all touchpoints with a client into a single chronological view. This includes:
Emails — sent and received emails linked to the company (via domain matching)
Tasks and projects — task completions, project launches, status changes
File activity — uploads, downloads, and shared files
Portal activity — client actions in the portal (task completions, file uploads, request submissions)
Status changes — company status updates, assignee changes
The timeline is accessible from the company record and provides a complete audit trail of the relationship. Use it when you need to reconstruct what happened with a client, verify when a document was shared, or confirm that a follow-up was sent.
How AI Insights Connect to Pulse
The Pulse health score (0–10) displayed on every company record draws from several data sources, and AI insights are a key input:
Email engagement — frequency and recency of email exchanges
Sentiment trend — whether the tone of communication is improving, stable, or declining
Task completion timeliness — whether tasks linked to the company are being completed on time or slipping
Manual adjustments — team members can manually adjust the Pulse score when they have context the AI doesn't capture
Companies with a Pulse score of 3 or below are labeled At Risk. Companies scoring 3–5 are labeled Poor. Both categories surface at the top of the default Companies list sort order to ensure at-risk relationships get attention.
Troubleshooting
AI summary shows "No data available." The company has insufficient email interaction history. Ensure your team has connected email in CX and that the company's website domain field is populated correctly for email matching.
Sentiment appears inaccurate. Sentiment analysis is based on email language patterns and may not capture context that exists outside email (phone calls, in-person meetings). Use it as a directional signal alongside your own relationship knowledge. If the score feels off, check whether the website domain is matching emails from the wrong company.
Pulse score seems too low despite a healthy relationship. Pulse weighs email engagement heavily. If your team communicates with a client primarily by phone or in person, the Pulse score may underweight the actual relationship health. Consider using the manual adjustment to bring the score in line with reality.
Related Articles
Companies Overview — understand how company records anchor client relationships in CX
Dashboards Overview — monitor Pulse scores and relationship health across your book of business
Inbox Overview — manage email communication that feeds AI insights
How CX Measures SLA Adherence — understand the metrics behind operational performance tracking