Genesys Cloud - Main

 View Only

Sign Up

  • 1.  Building More Accurate AI Summaries - Numbers

    Posted 09-29-2025 01:11

    Let's create a new model to produce more accurate Agent Copilot Summaries

    The new model will have to consist of Agent Copilot / AI Studio Summaries / Dialog Engine Bots Intent Miner (Intent Foundation) to produce the most accurate summaries available today.  All three having a symbiotic relationship that will include a agent's ability to maintain effective communication with customer to solicit accurate transcripts and summaries.

    Genesys Cloud – Symbiotic Model for Accurate Summarization

    Agent Copilot Intents & Utterances

    Agent Copilot supports live agents with real-time prompts and utterances to reinforce numeric accuracy.

    Member Verification

    - Please provide your Member ID one digit at a time.

    - Let me repeat your ID to confirm: one-two-three…

    - Can I verify your date of birth for accuracy?

    Numeric Confirmation

    - I heard $250, is that correct?

    - Your copay is $25 per visit, right?

    - You said 20 visits remaining, can I confirm that?

    Coverage Inquiry

    - How many visits are included in my plan?

    - What is my deductible amount?

    - What is my coinsurance percentage?

    Claim Resolution

    - What was the payment for claim 123456?

    - Can you check my claim from December 21st?

    - Why did I owe $240?

     Dialog Engine Bot Intents, Utterances, and Slots

    Verify Member ID

    Utterances:

    - My Member ID is 987654321.

    - ID number four two three nine eight one six five.

    Slots:

    - MemberID: Regex: ^[0-9]{9,12}$

    Provide Date of Service

    Utterances:

    - The date of service was September 29th, 2025.

    - My claim was on 12/21.

    Slots:

    - DateOfService: Format: YYYY-MM-DD

    Provide Dollar Amount

    Utterances:

    - It was $500.

    - The deductible is five hundred dollars.

    Slots:

    - AmountUSD: Regex: ^\$?\d+(,\d{3})*(\.\d{2})?$

    Provide Claim/Authorization Code

    Utterances:

    - Claim ID AB12345.

    - Authorization code is XYZ-67890.

    Slots:

    - ClaimID: Regex: ^[A-Z0-9-]{6,18}$

    Slot Types

    The following slot types provide standardized capture and validation rules:

    Slot Name

    Format

    Validation Rule

    Notes

    MemberID

    9–12 digits

    ^[0-9]{9,12}$

    Preserve leading zeros

    DateOfService

    YYYY-MM-DD

    Valid calendar date

    Convert spoken → ISO format

    AmountUSD

    Currency ($###.##)

    Regex for money

    Numeric + formatted string

    Percent

    0–100%

    ^(100|[0-9]{1,2})%$

    Coinsurance/copay %

    VisitCount

    Integer (0–365)

    Range validation

    Visits allowed/used/remaining

    ClaimID

    6–18 alphanum

    ^[A-Z0-9-]{6,18}$

    Uppercase, optional hyphen

    Phone

    10 digits (US)

    ^[2-9][0-9]{2}[2-9][0-9]{6}$

    Normalize to E.164 optional

    AI Studio Summaries Configuration

    Configuration settings enforce clarity, numeric accuracy, and compliance.

    Configuration Settings

    - Language: English (US)
    - Conciseness: Concise (production), Detailed (QA/training)
    - Format: Grouped bullet points
    - Avoid PII: On (summaries show 'Member ID verified')
    - Participant Labels: Internal=Agent, External=Customer

    Predefined Insights

    - Reason for Call
    - Resolution
    - Action Items

    Custom Insights

    - Verification: Member ID, DOB, ZIP (masked)
    - Financials: Deductible, Copay, Coinsurance, OOPM
    - Coverage Limits: Visits allowed/used/remaining
    - Claim Metadata: DOS, ClaimID, Payment amounts

    Sample Summary Template

    Reason for Call:
    - Member asked about physical therapy visits and MRI claim cost.

    Key Details:
    - Verification: Member ID verified (structured field stored)
    - Benefits: 30 visits/year; 10 used; 20 remaining; $25 copay
    - Claim: DOS 2025-09-29; $500 deductible ($300 met, $200 remaining); 20% coinsurance; $240 member responsibility

    Resolution:
    - Benefits explained, claim resolved, no further action required.


    Symbiotic Relationship

    Dialog Engine Bots capture validated numeric slots in the IVR. Agent Copilot reinforces numeric confirmation in real-time with guided utterances. AI Studio consumes transcripts and structured slots to produce compliant bullet-point summaries. Together, they form a symbiotic system that increases accuracy, reduces agent workload, and ensures compliance.


    #ConversationalAI(Bots,AgentAssist,etc.)

    ------------------------------
    Alex Slocum
    ------------------------------


  • 2.  RE: Building More Accurate AI Summaries - Numbers
    Best Answer

    Posted 09-29-2025 11:54

    Hello Alex,

    Sounds like a solid idea, but I would recommend raising this in the Genesys Cloud Product Ideas lab so that it can be considered by the product management team. Once you have created an idea, come back and post a link to it so that others in the community can vote on or add to the idea.



    ------------------------------
    Sam Jillard
    Online Community Manager/Moderator
    Genesys - Employees
    ------------------------------



  • 3.  RE: Building More Accurate AI Summaries - Numbers

    Posted 09-29-2025 12:38
    I did post an idea on AI Summarization available before call ends / wrap up code, so agent have chance to review and update or correct if needed.

    I have received great feedback from Support/CSM and Product Managers were there is marked improvement in AI Summary Accuracy