How Humana Uses IBM Watson to Handle 7,000+ Voice Calls Daily at One-Third the Cost

Humana replaced its legacy IVR system with an IBM Watson-based conversational voice agent that handles 7,000+ provider calls daily, completing inquiries in 2 minutes at one-third the previous cost.

Impact

~66% (1/3 cost)

Cost Reduction

90-95%

Accuracy Rate

7,000+

Daily Voice Calls Handled

~2 minutes

Inquiry Completion Time

Challenge

Legacy IVR system transferred over 60% of calls to human agents at high cost. Over one million provider calls monthly, with most callers bypassing the IVR for outsourced call centers.

Solution

Watson-based conversational voice assistant combining seven language models and two acoustic models with speech customization to understand caller intent and deliver eligibility, benefits, claims, and authorization information.

Tools & Technologies

What Leaders Say

Providers can now call into the Watson-based solution and complete an inquiry in about two minutes without waiting to reach a call center representative.

Sara Hines, Director of Provider Experience and Connectivity, Humana
Get the full story.

Sign up to read complete case studies, access detailed metrics, and unlock all use cases.

Full Story

Humana, one of America's leading health insurance companies, was spending heavily on its interactive voice response (IVR) system that transferred over 60% of calls to human agents. With more than one million provider calls monthly, most callers bypassed the IVR entirely, reaching outsourced call centers at significant cost. Administrative staff needed faster access to routine pre-service questions about member health plan benefits and eligibility.

Humana partnered with IBM to develop a Watson-based conversational voice assistant combining multiple Watson applications on IBM Cloud with watsonx Assistant for Voice on-premises. The solution uses seven language models and two acoustic models with speech customization to understand caller intent, verify access permissions, and deliver specific eligibility, benefits, claims, authorization, and referral information.

The results were impressive: the system achieves 90-95% sentence error rate accuracy and handles inquiries at approximately one-third the cost of the previous system. It processes over 7,000 voice calls from 120 providers per business day, with typical inquiry completion in about 2 minutes — compared to the previous method that required 7-page fax responses.

Similar Cases

GA
Giles AI
95%
medical research data extraction accuracy

Giles AI, a London-based healthcare AI startup, built its medical research assistant on Google Cloud using Vertex AI, Gemini Pro, and Document AI to help researchers extract structured insights from millions of scientific articles. The platform achieved 95% accuracy in data extraction, a 98% agreement rate with human researchers, and helped one clinical customer cut research task time by 85%.

HealthcareVAVertex AIGGemini
A
AstraZeneca
40%
developer velocity increase with github copilot

AstraZeneca, one of the world’s largest pharmaceutical companies, unified 5,000 developers and scientists onto GitHub Enterprise, automated CI/CD with GitHub Actions, and deployed GitHub Copilot — achieving a 40% increase in developer velocity in its pilot program and generating 9 to 10 additional hours of productive output per developer each week. With drug development timelines measured in decades, the company views even marginal acceleration as directly impacting patient outcomes.

HealthcareGEGitHub EnterpriseGCGitHub Copilot
ES
Epic Systems
Over 50%
claude code usage from non-developers

Epic Systems — the healthcare technology company behind MyChart, used by 195 million patients — deployed Claude Code across its entire workforce, not just engineers. Today, more than half of Claude Code usage at Epic comes from non-technical employees, including a pharmacist who built a fully interactive MyChart prototype without writing a single line of code.

HealthcareCCClaude CodeCClaude
UM
UChicago Medicine
100
data-driven marketing segments built

UChicago Medicine, a leading academic health system, leveraged Salesforce Data Cloud, Marketing Cloud, and Agentforce to unify patient data and deliver hyper-personalized outreach at scale. In just four months, the team built 100 data-driven marketing segments, achieving a 60% campaign conversion rate and full ROI in under a year. The organization is now deploying Agentforce Voice to autonomously handle millions of nonclinical patient inquiries annually.

HealthcareDCData CloudHCHealth Cloud
I
InpharmD
80%
data storage cost savings

InpharmD's AI assistant, Sherlock, leverages Pinecone's vector database to deliver fast, accurate drug information to healthcare professionals. By embedding 30 million medical documents into a RAG pipeline, InpharmD achieved 70% better query accuracy, 95x faster first response times, and 80% cost savings on data storage.

HealthcarePPineconeCCanopy
M
Medlitix
90% (70min to 6min)
review time reduction

Medlitix implemented UiPath medical record summarization with DeepRAG, cutting clinical review from 70 minutes to 6 minutes per case (90% faster) with 95% accuracy and $1.2M savings.

HealthcareUPUiPath Platform
IH
Intermountain Health
27% per appointment
note time reduction

Intermountain Health deployed Microsoft Dragon Copilot to 2,500+ clinicians, reducing time spent on notes by 27% per appointment and fighting clinician burnout with AI-generated clinical documentation.

HealthcareMDMicrosoft Dragon Copilot
CW
Camping World
40%
customer engagement increase

Camping World deployed IBM watsonx Assistant as a virtual agent named Arvee across all web properties, increasing customer engagement by 40% and improving agent efficiency by 33%.

RetailIWIBM watsonx AssistantICIBM Consulting