Daniella's Customer Brain

OneAI - Pipeline success - weekly meeting.

3 min read

OneAI - Pipeline success - weekly meeting.

Date: 2025-12-10 | Duration: 27.59000015258789 min | Account: pipeline-success

External attendees: bruna.santos@homecare.com, chris.b@pipelinesuccess.com, david@pipelinesuccess.com, ecorrons@homecare.com, jason.o@pipelinesuccess.com, ncuesta@homecare.com

Summary

The team discussed key AI call handling challenges, including a technical bug in voicemail detection confirmed by Amit Ben, with an expected fix to improve accuracy and reduce false activations. Jason O’Neill and Amit addressed the difficulties of handling calls with noise or strong accents, implementing platform-level triggers to manage low-quality calls, thus enhancing ticket creation. They also agreed to expand caller type classification to include an ‘Other’ option for better call routing. Ongoing issues with ticket creation consistency were acknowledged, and Amit committed to reviewing these failures. The team emphasized real-time troubleshooting through Slack for quick issue resolution, and they strategized a phased rollout for overnight call routing, focusing on maintaining service quality while gradually expanding AI capabilities.

Key Points

🔧 AI Call Handling Issues: Technical bug causes incorrect voicemail detection; fix due tomorrow improves accuracy and user experience. 📞 Noise Handling Solutions: New triggers will filter low-quality calls, enhancing caregiver interactions and ticket creation rates. 📋 Caller Type Expansion: Adding ‘Other’ caller type to streamline ticket creation for uncommon requests, reducing false positives. 🔍 Ticket Creation Consistency: Ongoing review of ticket failures; aim for reliable generation to enhance customer support metrics. 💬 Real-Time Collaboration: Issues handled via Slack for quick resolution, enhancing agility and reducing downtimes in AI operation. 📈 Strategic AI Scaling: Focusing on reducing false positives through improved classification, ensuring smoother user experiences and operational efficiency.

Overview

  • AI Call Handling Issues: Technical bug causes incorrect voicemail detection; fix due tomorrow improves accuracy and user experience.
  • Noise Handling Solutions: New triggers will filter low-quality calls, enhancing caregiver interactions and ticket creation rates.
  • Caller Type Expansion: Adding “Other” caller type to streamline ticket creation for uncommon requests, reducing false positives.
  • Ticket Creation Consistency: Ongoing review of ticket failures; aim for reliable generation to enhance customer support metrics.
  • Real-Time Collaboration: Issues handled via Slack for quick resolution, enhancing agility and reducing downtimes in AI operation.
  • Strategic AI Scaling: Focusing on reducing false positives through improved classification, ensuring smoother user experiences and operational efficiency.

Action Items

Amit Ben Investigate and fix the technical bug causing AI to incorrectly trigger speaking on behalf of a voicemail call (05:01) Enhance user interface to expose more detailed AI logs and decision-making data to the team (05:31) Develop platform-level automatic volume adjustment for AI during noisy calls to improve conversation clarity (11:30) Add “other” caller type option to call classification to handle non-caregiver/family callers correctly routing to generic ticket creation (16:20) Monitor and ensure no false positives in ‘other’ caller category; tighten classifier scope for Harmony AI accordingly (22:45) Collaborate offline with support team to resolve ticket creation failures and looping conversation issues impacting live calls (24:30)

Jason O’Neill Share examples of calls where tickets were verbally created by AI but not actually generated, specifically in the generic ticket flow (20:06) Continue testing new caller types and ticket routing during the week ahead (26:39)

Daniella Block Take note and track suggested platform improvements on noise detection and caller handling (12:58) Support monitoring of calls categorized as “other” to confirm accuracy and no false positives occur (22:50)

Noel Share specific examples of calls with multiple dispositions via Slack to Daniela and Roy for immediate issue resolution (25:58)

Roy Lahav Provide real-time support for resolving issues reported by Noel through Slack (26:00)

Graph View