Feat: Decision Interface and Feedback Loop, Next Actionable Step for Legal Team

[kofki]

🚀 Goal

Create a "Tinder-style" interface for the legal team to quickly process AI-generated actions. Every decision (Accept/Reject) should train the system to be more accurate over time.

---

🛠 Tasks

• Gesture Interface: Implement a swipe-able card UI for legal recommendations.

• Swipe Right: Accept and trigger the "Next Action."
• Swipe Left: Reject and trigger a feedback modal.

• Feedback Mechanism: Create a pop-up for rejected actions to capture why (e.g., "Incorrect Statute," "Missing Context").
• Next Action Trigger: Automated workflows based on acceptance (e.g., auto-drafting a response or filing a document).
• Learning Loop: Log all decisions into MongoDB to refine future routing logic.

---

###📝 Logic Flow

• AI Suggestion: System presents a legal strategy card.
• User Interaction:

• If Accept: Move to status: "executed".
• If Reject: Move to status: "needs_human_input"

• User Interaction:: Save the choice and the "Reason Code" to the feedback collection.

---

✅ Acceptance Criteria

• Users can swipe through a stack of recommendation cards.
• Swiping Right successfully triggers the next step in the legal workflow.
• Swiping Left forces a mandatory text box for "Reason for Rejection."
• The database stores the user ID, the recommendation ID, and the outcome.
• The interface updates in real-time without needing a page refresh.

---

🏁 Pull Request Requirements

• UX/UI: Ensure the swipe animation is smooth and intuitive (it can be with mouse or with keys)
• State Management: Correctly handle the "Empty State" when no more recommendations are left in the queue.
• Database: Verify that the feedback data is being saved in a format ready for future model retraining.