AI Intent Detection Workflow Action in Sparkforge

The AI Intent Detection action analyzes text and routes the workflow down one of three branches based on detected sentiment. It replaces legacy intent detection inside IF/Else conditions and is more reliable for triggering automated responses on customer messages, survey replies, or form submissions.

What it does

Drop the action into any workflow. It reads the text you point at, decides whether the sentiment is positive, negative, or neutral/unclear, and creates three branches automatically:

• POSITIVE — satisfaction, agreement, praise.
• NEGATIVE — dissatisfaction, complaints, disagreement.
• NONE — neutral or low-confidence text.

Set up AI Intent Detection

Step 1: Add the action

In the workflow builder, search the actions menu for AI Intent Detection and drop it where you want sentiment to be analyzed.

Step 2: Configure inputs

• Action Name — descriptive label, e.g. Analyze Feedback Sentiment.
• Input Text — either a static value or a dynamic variable. Use {{message.body}} for incoming messages, or pick any custom value that holds text.

Step 3: Build branch actions

The action auto-creates the three branches. Wire each one with the response you want.

• POSITIVE — send a thank-you, request a review, tag the contact as a happy customer.
• NEGATIVE — escalate to support, send a recovery offer, schedule a follow-up call.
• NONE — request clarification or route to manual review.

Example: feedback survey routing

After a customer submits a feedback form, run the response through AI Intent Detection. A POSITIVE response triggers a Google review request. A NEGATIVE response opens a support ticket and texts the owner. A NONE response sends a clarifying question.

Tips

• Use this on any text field, not just chat — survey answers, form fields, and email replies all work.
• Combine with tags so you can segment contacts by sentiment over time.
• Keep NEGATIVE-branch actions fast — recovery responses lose impact after a few hours.