To help us better level set our own expectations, roughly how long do you find it takes for the system to improve its accuracy? We've been toying with the Custom Dictionary for a few months now, and are still seeing it misspell our company name (for starters). Ultimately it sounds like we would have to devote an FTE to take the feedback from our QA Teams (who look at the transcripts far more) or comb through recordings on a routine basis to keep up/stay ahead of things--all in an effort to tighten up the accuracy.
For example, a few days ago we did a little cleanup in our Dictionary and had already employed the recommendations you made above. Yet today we're still seeing the misspelling of our company name, or lack of capturing it altogether (despite the agents speaking it clearly.) It's almost as if for every "sounds like" word we add (e.g. we added each of those "sounds like" words incrementally, not all at once), the system finds a different word to use instead of the term we want it to, haha.
Original Message:
Sent: 05-29-2025 20:19
From: Jose Ruiz
Subject: #WEMay | What's on Your Mind About AI Scoring & Copilot?
Hello @Gene Gutierrez and @Brian Jones
Thank you for the thoughtful and detailed feedback - you've highlighted a crucial aspect of delivering high-performing AI-driven features like Virtual Supervisor and Supervisor Copilot: high-quality transcription. We completely agree that accurate speech-to-text output is foundational to ensuring reliable interaction scoring, sentiment analysis, real-time translation, and AI-generated summaries.
To address the specific challenges you've raised, I'd like to share more about the Custom Dictionary feature, which is designed to improve transcription accuracy by tuning the model to recognize key business-specific terms - including organizational names, product terminology, and industry jargon.
With Custom Dictionary, you can:
-
Add custom phrases (e.g., "Thank you for calling Presbyterian") to help the model understand and prioritize relevant business language.
-
Improve recognition of difficult or misheard terms by adding "sounds like" alternatives. This is particularly helpful when commonly misrecognized phrases yield similar-sounding but incorrect words, as in your example.
We recommend starting with boost-only entries for important terms and then iteratively enhancing them with "sounds like" variants for any recurring misrecognitions - as long as they don't conflict with valid words in the language.
This feature effectively gives you a way to embed business context directly into the transcription model, improving accuracy over time, especially in areas most important to your organization.
I'd be happy to connect directly to review your current transcription challenges and provide tailored recommendations on Custom Dictionary usage and other tuning options. Together, we can ensure that transcription-dependent features are delivering maximum value and reliability for your team.
Please don't hesitate to reach out!
------------------------------
Jose Ruiz
Genesys - Employees
Product Manager
jose.ruiz@genesys.com
Original Message:
Sent: 05-28-2025 12:14
From: Brian Jones
Subject: #WEMay | What's on Your Mind About AI Scoring & Copilot?
Totally agree with this feedback--even down to the organization name misspelling ( Genesys thinks "Ascension" is "extension" or similar, despite the continued Dictionary/Phrase Management.)
With Genesys' increased focus on AI capabilities, I would love to hear Genesys' native transcription service improvement plan given that must be lockstep if there's any chance for their AI to keep up with the pack.
------------------------------
Brian T. Jones | Ascension | Senior Specialist - Technology
Original Message:
Sent: 05-23-2025 16:18
From: Gene Gutierrez
Subject: #WEMay | What's on Your Mind About AI Scoring & Copilot?
While your new Virtual Supervisor and Supervisor Copilot features represent a significant leap forward in contact center AI, their effectiveness is fundamentally linked to the underlying voice-to-text transcription quality. Our experience suggests that there's a valuable opportunity to refine the current transcription model to fully support the advanced capabilities these features promise.
We've observed instances where transcriptions, even of common phrases including our organization's name, don't consistently capture the spoken word accurately. For example, "Thank you for calling Presbyterian" has sometimes appeared as "Thank you for calling Cer," "customer", or "presents". These observations suggest an area for further optimization.
This level of transcription accuracy has an impact on the reliability of both Virtual Supervisor and Supervisor Copilot:
Virtual Supervisor
The precision of automated interaction scoring and sentiment analysis is directly tied to the accuracy of the underlying text. If the AI doesn't consistently capture what was said, the insights and assessments it provides can be less reliable, potentially affecting the effectiveness of automated quality management.
Supervisor Copilot
The utility of features like On-Demand Translation and AI Insights relies heavily on precise source transcription. If the original speech isn't accurately converted to text, real-time translations may be inaccurate, and AI-generated summaries of contact reasons, resolutions, or follow-up actions might not fully reflect the conversation. This could mean agents or other users would need to spend additional time verifying information, which can reduce the intended efficiency gains. Furthermore, the system's ability to interpret context, such as an internal transfer leading to "I don't know" responses, becomes more complex if the foundational transcription isn't precise.
Recommendations for Optimization
To help Virtual Supervisor and Supervisor Copilot achieve their full potential, we suggest focusing on enhancements to the voice-to-text transcription engine. We recommend:
- To fully realize the potential of Virtual Supervisor and Supervisor Copilot, we believe there's an opportunity to further refine the foundational speech-to-text accuracy. This refinement will directly enhance the reliability and actionable nature of the AI insights these features aim to deliver.
- Exploring capabilities to train the transcription model on client-specific vocabulary would be highly beneficial, ensuring accurate recognition of organizational names, product terms, and industry jargon.
- Implementing greater transparency around transcription confidence levels could allow supervisors to identify and prioritize manual review for less certain transcriptions. Additionally, robust feedback mechanisms for correcting errors could help continuously improve the model's learning.
Enhancing this core transcription capability will build even greater trust in the platform's AI features and allow us to fully leverage the power of Virtual Supervisor and Supervisor Copilot.
------------------------------
Gene Gutierrez | Workforce Analyst
Presbyterian Workforce Management
Albuquerque, NM
Original Message:
Sent: 05-21-2025 05:00
From: Tracy Vickers
Subject: #WEMay | What's on Your Mind About AI Scoring & Copilot?
Have questions about AI Scoring, AI Translate, or AI Summary and Insights since the launch of Virtual Supervisor and Supervisor Copilot?
We know there's a lot to explore-what these features are, how they work, and most importantly, how they can benefit you.
To help you stay informed, there is a helpful Resource Centre page that includes detailed information and a list of FAQs:
🔗 About Virtual Supervisor and Supervisor Copilot
Can't find the answer you're looking for? Feel free to ask your question right here-we're here to help!
#WEMay
------------------------------
Tracy
Genesys
------------------------------