Follow-up Intelligence

AI Task Suggestions

AI Task Suggestions detect promises, commitments, next steps, and due dates from communication so the team does not rely on memory or manual task entry.

1. Feature Overview

During calls, texts, and emails, agents and prospects often mention future actions. AI Task Suggestions converts these natural language commitments into structured task drafts that can be approved and saved.

Detect commitments

Find “I’ll call tomorrow,” “send pricing,” “follow up next week,” and similar promises.

Create task drafts

Suggest task title, due date, priority, assignee, and linked record.

Keep users in control

Agent approves, edits, or dismisses before task creation.

2. Task Suggestion Workflow

flowchart TD
 A[Communication / Call Summary] --> B[Extract Commitments]
 B --> C{Actionable?}
 C -->|No| D[No Task Suggested]
 C -->|Yes| E[Parse Task Fields]
 E --> F[Title]
 E --> G[Due Date]
 E --> H[Priority]
 E --> I[Assignee]
 E --> J[Linked Guest Card]
 F --> K[Task Draft]
 G --> K
 H --> K
 I --> K
 J --> K
 K --> L{User Decision}
 L -->|Approve| M[Create Task]
 L -->|Edit| N[Update Draft then Create]
 L -->|Dismiss| O[Log Dismissal]

3. What the AI Should Detect

Pattern	Example	Task Suggestion
Agent commitment	“I’ll send pricing tomorrow.”	Send pricing guide tomorrow.
Prospect request	“Can you call me next week?”	Call prospect next week.
Document follow-up	“I’ll bring medical records Tuesday.”	Collect documents during Tuesday appointment.
Referrer follow-up	“Email me the availability by 4 PM.”	Email availability before 4 PM.
Tour preparation	“She wants a quiet room near the garden.”	Prepare quiet/garden-side option for tour.

4. Detailed User Cases

User Case 1: Pricing follow-up commitment

Scenario: Agent texts, “I’ll send you the pricing guide tomorrow morning.”

AI behavior: Suggests task: “Send pricing guide,” due tomorrow morning, assigned to current agent, linked to guest card.

Result: The agent does not need to manually create a reminder.

User Case 2: Home care staffing check

Scenario: Agent tells referrer, “I’ll check overnight staffing and email you by 4 PM.”

AI behavior: Suggests high-priority task due 3:30 PM: “Confirm overnight staffing and email referrer.”

Result: The commitment is tracked before the deadline.

User Case 3: Tour preparation

Scenario: A daughter says her father needs a first-floor room and dislikes loud areas.

AI behavior: Suggests a task before the tour: “Prepare first-floor quiet room options.”

Result: The tour is better personalized.

5. Proposed User Experience

Inline suggestion

Task draft appears below the communication or summary with Approve, Edit, and Dismiss actions.

Task drawer

A side panel shows all AI-suggested tasks for the current guest card.

Morning Brief

Unapproved high-value task suggestions can appear in the agent’s daily brief.

Audit record

System stores whether the task was AI-suggested, edited, approved, or dismissed.

6. Data Model

Field	Purpose
suggested_title	Task title generated by AI.
due_date	Parsed natural language date/time.
priority	Low, medium, high, urgent.
assignee_id	User suggested to own the task.
source_record_id	Message, call, summary, or activity that caused the suggestion.
approval_status	Pending, approved, edited, dismissed.

7. Acceptance Criteria

System can detect actionable follow-up commitments from SMS/email/call summaries.
Suggested task includes title, due date if available, assignee, linked record, and priority.
User can approve, edit, or dismiss task suggestion.
Approved suggestions create real CRM tasks.
Dismissals are logged to improve future suggestions.
AI task suggestions appear in guest card and Morning Brief where appropriate.

8. Success Metrics

Task approval rate

How many AI suggestions become real tasks.

Follow-up completion

Improvement in completed follow-up tasks.

Missed commitment reduction

Reduction in promised but untracked follow-ups.

Edit rate

How often AI task drafts need changes.