ability to Train AI on Past Chat History to Improve Support Accuracy and Personalization
Migrated from Canny.
Canny Post ID: 680a954bb78da8b513514376
Canny URL: https://spur.canny.io/admin/board/feature-requests/p/ability-to-train-ai-on-past-chat-history-to-improve-support-accuracy-and-persona
Category: New Feature
Original status: open
Original votes: 1
Original comments: 0
Original author: Khizar Arif
Created at: 2025-04-24T19:47:23.406Z
Original request details:
It would be incredibly useful if Spur’s AI could be trained directly on our past customer chats — for example, from Jan 2024 to April 2025 — especially those on our support number.
Instead of relying solely on help center articles or manually fed data, the AI would already “know” what worked in the past. It could automatically reference resolved edge cases, common solutions, and even handle unusual queries by recognizing how they were previously handled.
Beyond problem-solving, the AI could use sentiment analysis on past chats to learn how to deal with different customer tones and behaviors. This would lead to more empathetic, tailored responses and faster resolutions — especially for returning customers.
Why it’s valuable:
• Removes the need to recreate support content the AI already “lived through”
• Improves accuracy for edge cases and uncommon issues
• Reduces onboarding time for the AI
• Makes AI responses more nuanced and human-like by understanding tone/context
Would love to see this as a premium or AI Max feature. Let me know if others feel the same — upvote if you’d use this!
1 Comment
Sign in to comment
REQUEST_SUMMARY_V2 Problem - It would be incredibly useful if Spur’s AI could be trained directly on our past customer chats — for example, from Jan 2024 to April 2025 — especially those on our support number. Instead of relying solely on help center articles or manually fed data, the AI would already “know” what worked in the past. It could automatically reference resolved edge cases, common solutions, and even handle unusual queries by recognizing how they were previously handled. Beyond problem-solving, the AI could use sentiment analysis on past chats to learn how to deal with different customer tones and behaviors. This would lead to more empathetic, tailored responses and faster resolutions — especially for retu Expected behavior - Support this behavior in shared inbox without manual workaround - Maintain clear ownership and state transitions for conversations - Keep agent actions auditable in activity logs Business impact - Reduces manual operational effort for teams - Improves consistency and customer experience - Helps teams move faster with fewer support escalations Legacy category: New Feature
Sign in to comment