Workforce Engagement Management

Hi everyone,
Thank you to everyone who joined our May 2026 WEM Advisory Webinar | Orchestrating Your Quality Processes with AI. We received some excellent questions during the session, so we have grouped the answered questions below by topic and lightly edited them for clarity.

You will find the recording here:  https://community.genesys.com/viewdocument/recording-wem-advisory-webinars-series-orchestrating-your-quality-processes-with-ai?CommunityKey=f6e20f73-2c94-4b44-be95-118216aafb4f&tab=librarydocuments

Note: Roadmap and availability comments reflect what was shared during the webinar. Please check the linked resources or Ideas Portal items for the latest status before relying on delivery timing.

Helpful resources shared during the webinar

- AI Scoring best practices: https://help.genesys.cloud/articles/ai-scoring-best-practices/
- Generate an agent auto-complete evaluation: https://help.genesys.cloud/faqs/how-do-i-generate-an-agent-auto-complete-evaluation-2/
- Agreement rate for auto-complete evaluation forms: https://help.genesys.cloud/faqs/what-is-the-agreement-rate-and-how-is-it-used-to-fine-tune-an-auto-complete-evaluation-form/
- View an interaction’s customer journey: https://help.genesys.cloud/articles/view-an-interactions-customer-journey/
- Idea: Quality form export/download: https://genesyscloud.ideas.aha.io/ideas/WEQUAL-I-281
- Idea: Additional quality-manager review/audit workflows: https://genesyscloud.ideas.aha.io/ideas/WEQUAL-I-475
- Idea: AI summary language support for Arabic interactions: https://genesyscloud.ideas.aha.io/ideas/DARSTA-I-458

Q&A

AI scoring setup and prompt design

**Q: Can policies automatically select calls for quality evaluation scoring? Can we use multiple forms and send them to multiple evaluators?
**A: Yes. You can create multiple policies. Each policy uses one form.

**Q: Does AI scoring require topics and phrases to be configured?
**A: No. AI scoring uses the evaluation question and help text as the prompt, then uses the transcript to answer.

**Q: Evaluation Assistance requires speech analytics topics to be fully configured, correct? 
**A:** Correct. Evaluation Assistance relies on configured speech analytics topics.

**Q: Where does AI get the information that something specific is required, such as in a compliance question? Is that submitted in the form, or do we add those policies somewhere else?
**A: AI uses the prompt/help text you provide at the question level. Use the help text to define the exact requirements, expected evidence, and scoring criteria.

**Q: Can AI scoring support questions that are not simple yes/no questions, such as “below expectations,” “meets expectations,” or “unacceptable”?
**A: Yes. This depends on the question type selected for each question.

**Q: Can AI scoring give partial credit if the agent completed part of the requirement? 
**A: Yes. You can use three answer choices, with five answer choices expected in a future enhancement. Make sure the help text clearly defines what qualifies for full credit, partial credit, and no credit.

**Q: Can we mix traditional evaluations and AI scoring?
**A: Yes. This is a strong way to use both approaches for what they do best.

**Q: Can AI scoring determine whether an agent offered an alternative option when the customer’s requested option was unavailable? 
**A: Yes. Make the attribute and scoring expectation clear in the question and help text so AI knows what evidence to look for in the transcript.

**Q: If we correct an AI answer, will the AI learn from that correction? 
**A: No. Corrections do not train the AI. A correction usually indicates that you may need to adjust the question or help text.

Time thresholds, metadata, and transcript-based scoring

**Q: Can AI scoring isolate the agent and score whether the agent greeted the customer within a 10-second threshold? 
**A: AI scoring does not use interaction metadata to measure a 10-second threshold. To evaluate whether the agent greeted the customer promptly, frame the question around transcript evidence instead. For example: “Did the agent greet the customer before the customer had to prompt them?” Help text can clarify that a “Yes” means the agent greeted before any customer prompt such as “hello” or “are you there,” and a “No” means the customer prompted before the agent greeted or no greeting was provided.

**Q: Can we add a form question that asks whether the agent kept the customer on hold for under a set number of minutes? 
**A: AI scoring is not using metadata to evaluate questions. Many customers use hold-time KPIs in gamification rather than measuring this on QM forms. AI can evaluate transcript-based evidence, such as the customer saying they have been holding too long, if that appears in the transcript.

**Q: Can AI use CIS or external data as context so it knows when certain conditions apply?
**A: Genesys Cloud AI can use external data, including CIS data, as context for AI scoring and interaction analysis, although the primary direct scoring mechanism relies heavily on transcript analysis.

**Q: If AI scoring uses the transcript, could an incorrect transcript lead to an incorrect score? 
**A: Yes, AI scoring is based on the transcript, so transcription accuracy matters. Continued model improvements and Dictionary Management can help reduce this risk.

**Q: Can the evaluation form using Evaluation Assistance or AI differentiate between the IVR/automated system and the representative/customer conversation? 
**A: You can use the option to suppress recording when the caller is interacting with the IVR. If that part is not recorded, questions will not be answered based on the IVR portion of the conversation.

Automated evaluations and quality policies

**Q: How can Genesys score 100% of interactions without a person creating the evaluation?
**A: Create an Agent Auto Evaluation form, then set up the scoring rules in Programs under Speech and Text Analytics.

**Q: Is anything needed beyond creating a program to set up AI scoring? 
**A: Yes. Create the Agent Auto Evaluation form, then configure the scoring rules in the Program.

**Q: Can different lines of business have different forms? 
**A: Yes. For fully automated evaluations, you would typically use a Program for each line of business.

**Q: Can auto-evaluations avoid affecting agent performance while still using weighted questions? 
**A: Auto-evaluations currently are not released to the agent. The score is recorded, but it is not included in Agent Evaluation reports.

**Q: Can automated evaluations be configured to run only when certain rule-based conditions are met? 
**A: AI scoring runs only when a conversation meets at least one of these minimum criteria: the conversation lasts 30 seconds or longer, or the transcript contains at least 300 characters. Within policies, you can also select thresholds such as agent connected duration.

**Q: Is anything on the roadmap to support additional selection conditions, such as call duration or wrap-up code? 
**A: Yes. A roadmap item was shared to allow more selective interaction criteria, such as topics, sentiment, direction, and wrap-up code.

**Q: Can it be set to audit calls only based on the wrap-up code chosen? 
**A: This was shared as a roadmap item, along with more selective criteria such as sentiment, topics, and direction.

**Q: Is there a way within quality policies to select interactions with screen recording? 
**A: Yes. You can select the screen recording option in the evaluation policy so selected interactions also have screen recordings.

Evaluation Assistance, defaults, and answer choices

**Q: When will Evaluation Assistance answer both choices under a question? Currently we have to choose whether it pre-fills “Yes” rather than “No.”
**A: When building the question, set your default answer to the other option. Then, if Evaluation Assistance does not answer, the default answer will be used.

**Q: When setting up an Evaluation Assistance question, what is the purpose of a default answer for yes/no questions? 
**A: If you are using the Auto Agent Evaluation form type, each question must have a default answer. For Evaluation Assistance, the condition is placed on one answer, so a default answer allows the question to be answered if the configured topic is not tagged.

Reporting, AI insights, and APIs

**Q: How close are we to out-of-the-box reporting for Auto-Evaluation forms? 
**A: AI scoring reporting combines automated evaluation results with layered analytics and insights, rather than existing as a single standalone report. It includes automated scores, explainable outputs such as reasons and confidence, agreement reporting, and traditional QA reporting enhanced by AI-driven data. Question-level and aggregate performance reporting were discussed as roadmap areas.

**Q: Is there reporting available for Agent Auto Scoring in one place, or do we need to open interactions individually? 
**A: For now, use the Interactions view. Add a filter for the Auto Form, then add the Score and Critical Score columns. Evaluation data can also be pulled via APIs.

**Q: Can auto-evaluation results be fetched for internal dashboards? 
**A: Yes. The data exists on the interaction and can be pulled via APIs.

**Q: Is reporting available for AI Insights, such as common contact reasons or recurring low sentiment? 
**A: Enhancements are on the roadmap. For now, AI insights and summaries can be pulled through APIs and analyzed outside the UI. The feature was originally designed to simplify supervisor, leader, and coach workflows rather than act as a reporting dataset.

**Q: Can we get AI analysis by individual agent across all evaluations in a period, or only call by call? 
**A: Form, group, and question-level reporting was shared as a roadmap item for Q3.

**Q: Will there be reporting on specific questions across teams or individuals?
**A: This was shared as being on the current roadmap.

**Q: Is Agreement Scoring available? 
**A: Yes. Agreement Scoring is available now.

**Q: Is the agreement score for both AI-scored questions and AI-assisted questions? 
**A: Agreement score is for AI questions.

**Q: AI Insights references 50 conversations/tokens. Is there a limit within a conversation, such as a conversation longer than 30 minutes? 
**A: Each conversation is processed as long as it fits within the transcript size limit of approximately 5,000 tokens. There is no cap on the number of insights generated within that conversation.

Language, summaries, and sentiment

**Q: Does AI scoring translate and summarize, or does it only work in English? 
**A: You can draft AI-scored questions in one language and still have them answered if the transcript is in a different language. For example, a form written in English can still be completed and answered when the transcript is in Spanish.

**Q: Can AI summaries be generated in English for Arabic interactions? 
**A: A related idea has been submitted in the Ideas Portal: https://genesyscloud.ideas.aha.io/ideas/DARSTA-I-458

**Q: Can we edit what is considered negative sentiment? For example, the word “cancel” does not always mean the sentiment is negative.
**A: Genesys Cloud uses the context of the transcript, not only individual keywords. You can provide sentiment feedback in the UI if the result does not align with your business context. Navigate to Menu > Conversation Intelligence > Speech and Text Analytics > Sentiment Feedback.

Agent self-evaluation, disputes, and audit trail

**Q: Are agents able to complete their own self-assessments, or can they only review scores and dispute them? 
**A: You can provide agents with permissions to complete evaluation forms.

**Q: Can an employee listen to and evaluate only their own interactions? 
**A: Yes, provided the employee has the required permissions.

**Q: Is the maximum number of disputes counted per question, per form, or per evaluated user?
**A: The maximum number of disputes is per evaluation. The limit shared was five disputes per individual evaluation.

**Q: Can a senior quality manager score an evaluation completed by a junior quality manager for feedback or audit purposes? 
**A: There is an Idea submitted for this workflow: https://genesyscloud.ideas.aha.io/ideas/WEQUAL-I-475

**Q: What are Audit Trail and Customer Journey used for on the interaction? 
**A: Audit Trail shows who viewed the interaction, who viewed or completed an evaluation, and related activity. Customer Journey details can be reviewed here: https://help.genesys.cloud/articles/view-an-interactions-customer-journey/

Multi-agent, transfers, and supervisory support

**Q: How will hands-off auto-evaluation work for message interactions handled by multiple agents? How will score assignment and not-applicable questions work for each agent?
**A: An enhancement was shared for the end of May to allow scoring for each agent who touches an interaction.

**Q: During transfers, can topics be specified by agent? 
**A: This was shared as being on the roadmap.

**Q: Can AI identify when supervisory support is needed and alert a supervisor? 
**A: This can be done using speech and text topics, triggers, and workflows.

Forms and roadmap items

**Q: Is there a timeline for increasing the form limit to 50 questions? 
**A: This was shared as being in scoping/development, with no exact release date announced.

**Q: Is there anything on the roadmap for downloading quality forms? We need to upload forms into a separate system. 
**A: There is an Ideas Portal item for this request. Product Management responded on 15 April 2026 that the request aligns with the roadmap and customer needs, development planning is underway, and updates will be provided as progress is made: https://genesyscloud.ideas.aha.io/ideas/WEQUAL-I-281

Thanks again to everyone who joined and contributed questions. Please continue the conversation in the WEM Community if you have additional questions or would like to share how your organization is approaching AI-assisted quality management.