← Back to Guides
🎯 AI Workflows & Memory: Closing the Loop
Qualtrics Summit X4 | March 18, 2026 | ~41 minutes
📄 Download Full Transcript (Original Text)
Presenters
- Satish — Qualtrics Presenter
- Priti — Qualtrics Presenter
- Cya — Live Demo Presenter
Core Theme: From Scripted to Intuitive Automation
Moving from deterministic workflows to AI-powered automations that understand context, memory, and can handle ambiguous situations.
Three Altitudes of Automation
| Level | Description |
|---|
| Predictive | Traditional workflows — if A then B, deterministic paths |
| AI-Assisted | Workflow with injected AI task to process unstructured data |
| Agentic | Goal-driven agent with tools — figures out reasoning itself |
The Memory Foundation (X ID)
Three steps to building memory:
- Ingest — Get data from multiple sources (email, calls, surveys)
- Enrich — Connect dots, link to individual via Experience ID
- Action — Surface in dashboards, workflows, or agents
"The memory itself is linked up to your Experience ID, and that's how we know all this data belongs to a specific individual."
AI Response Task Capabilities
Three main capabilities:
1. Cognitive Processing
- Summarize large volumes of feedback
- Multi-language translations
- Extract key issues and themes
2. Intelligent Routing
- Categorize issues automatically
- Assign priority
- Route to right team
3. Content Generation
- Draft personalized responses
- Send follow-up messages
- Human in the loop for approval
Real-World Use Cases
Contact Centers / Banking
- Summarize customer complaints
- Draft empathetic responses
- Identify urgent issues
- Auto-create and route tickets
Market Research
- Process large volumes of open-ended feedback
- Identify key themes and trends
- Extract product insights
Education
- Analyze student feedback at scale
- Identify student concerns and trends
- Personalized career advising
Key Metrics
- 8,800%+ growth in brands using AI response task since November launch
- 1 billion+ workflow executions per month
Demo Highlights
- Survey response triggers workflow via Experience Agent
- Workflow queries X ID for user history/preferences
- AI response task processes context, determines urgency
- Creates ticket with full context for live agent
- Ticketing agent uses X ID to recommend goodwill offers
"This is what's stopping us from making the user feel like they're talking to a brick wall — we have all that context captured in the ticket itself."
Alan's Questions (from Q&A)
"When the demo you gave us — that was a Qualtrics survey... behind the scene there's a survey flow that called a workflow triggered by a call to an agent, right?"
Answer: Experience agents connect to workflows as tools. Survey triggers agent → agent passes to workflow → workflow does heavy lifting.
"How is X ID connected to the activity? How are they recognized? Does the person have a called number, is a member, has purchase history?"
Answer: Survey must be non-anonymous (directed to individual) to get contact ID, which queries directory for attributes. Works with in-app surveys too.
"Can it be real time? Or do we have to wait for response to be submitted?"
Answer: Real-time processing available with AI response task. Requires AI SKU.
Availability
- AI Response Task: GA (generally available)
- Requirement: AI SKU license
- Experience Agents: Preview (TSM can enable)
Guide created: March 18, 2026 | Source: Note_20260318_1430_otter.ai