CVS Health
conversation intelligence and AI agents on 100% of contact center calls
CVS Health, which serves 185 million people a year, went from scoring 5% to 100% of its calls by deploying Cresta's conversation intelligence with a predictive CSAT on every call, and is extending Salesforce Agentforce Health AI agents to its Aetna and Caremark members.
Key points
- Conversation intelligence applied to 100% of member contact center calls.
- Cresta platform (summarization, predictive CSAT) plus Salesforce Agentforce Health agents.
- Share of scored calls raised from 5% to 100%, across 185 million people served a year.
- Evidence level B, status confirmed.
Objective
Move from a sample of manually scored calls to reading 100% of contact center interactions, to measure satisfaction on every call, reduce advisors' after-call work, and detect issues in real time rather than after weeks of surveys.
The deployment
CVS Health deployed Cresta's conversation intelligence in its member contact center. The company used to score 5% of its calls; it now scores 100%. The starting point was automatic call summarization, which reduced advisors' after-call work, before opening up a predictive satisfaction score (CSAT) on every call. CVS Health serves 185 million people a year. In parallel, the company is extending a layer of AI agents via Salesforce Agentforce Health for its Aetna and CVS Caremark businesses, across a scope of several tens of millions of members.
Results Proof B
The central result (going from 5% to 100% of scored calls, predictive CSAT) is documented by a quantified platform case study (Cresta), corroborated by a CVS/Salesforce announcement on the deployment of AI agents at the scale of several tens of millions of members. Quantified vendor case study = B.
How it works
Documented architectureThe stack in detail
- plateforme Cresta contact center conversation intelligence: transcription, automatic call summarization, and predictive CSAT on 100% of calls
- plateforme Salesforce Agentforce Health contact center AI agents for Aetna and CVS Caremark members, extension announced in 2026
- outil Transcription vocale temps reel speech-to-text component integrated into the conversation platform, a prerequisite for summarization and scoring
How it runs, concretely
For ops teams-
1Capture and transcription AI
Every member call is transcribed and made analyzable.
-
2Automatic summarization AI
The system generates a call summary that lightens the advisor's after-call work.
-
3Predictive scoring AI
A predictive CSAT is computed on 100% of calls, with a per-advisor view.
-
4Detection and remediation Customer experience team
The teams spot issues in real time and act without waiting for a survey.
-
5AI agent extension AI and member relations team
AI agents (Agentforce Health) assist Aetna and Caremark advisors with members.
The content of the transcribed calls and the predictive CSAT computed on it. Without reliable transcription or a consent basis for call analysis, coverage falls back to the manual sample it started from.
How your customers perceive this type of use
Sourced studiesLes consommateurs n'acceptent pas les chatbots par defaut : 64% prefereraient que les entreprises n'utilisent pas d'IA dans leur service client (Gartner, 2024) et pres d'un utilisateur sur cinq du service client par IA n'en retire aucun benefice (Qualtrics, 2025). L'acceptation se construit sur trois conditions mesurees par Salesforce : savoir qu'on parle a une IA, pouvoir escalader vers un humain, comprendre la logique de l'agent.
Acceptance conditions
- Etre informe qu'on parle a une IA et non a un humain (pres de 75% le demandent, Salesforce 2024)
- Un chemin d'escalade clair vers un agent humain (45% plus enclins a utiliser l'agent IA, Salesforce 2024)
- Une logique de l'agent clairement expliquee (44% plus enclins, Salesforce 2024)
Red lines
- Rendre l'humain injoignable : c'est la premiere inquietude des consommateurs sur l'IA dans le service client (Gartner 2024) et 50% craignent que l'IA les coupe du contact humain (Qualtrics 2025)
- Remplacer le service client par l'IA sans alternative : 53% envisageraient de partir chez un concurrent (Gartner 2024)
Sources: Salesforce 2024 · Gartner 2024 · Qualtrics 2025
How to replicate
Inference, not sourcedData prerequisites
- transcribed calls
- CSAT history to train the scoring
- a consent basis for call analysis
Org prerequisites
- contact center with call volume
- quality/customer experience team
- a compliance framework for recording
Possible stack
- conversation intelligence platform (Cresta, Observe.ai, etc.)
- AI service agents (Agentforce, custom)
- voice transcription
The plan, step by step
- Step 1Set the compliance framework (call recording, consent, health data) and contract the platform.Deliverable: Validated compliance file and signed contract
- Step 2Wire transcription onto the calls and launch automatic summarization on a pilot team.Deliverable: Summaries in limited production, after-call time measured before / after
- Step 3Generalize automatic summarization and calibrate the predictive CSAT on the history of manual scores.Deliverable: Predictive scoring validated against the manually scored sample
- Step 4Move to 100% of calls scored with dashboards per team and per advisor.Deliverable: Full call coverage with management views
- Step 5Set up the real-time remediation loop on the customer experience side, without waiting for surveys.Deliverable: Documented and operated detection-action process
First step: Wire up call transcription and launch automatic summarization as a first, fast-ROI use case.
The same pattern, proven in other industries
ADT
genAI customer agent
Air India
genAI customer agent (automated customer service)
AT&T
Governed enterprise genAI assistant: an LLM connected to internal knowledge, open to tens of thousands of employees and called at scale by automated systems
Sources
- S1 How CVS Health Went From Scoring 5% of Calls to 100% with AI-powered Conversation Intelligence Interested party archive pending
- S2 CVS Health to Deliver Faster, More Personalized Call Center Care for Millions of Members with Salesforce's Agentforce Health Interested party archive pending
An error, newer info, a source?
This page lives on its accuracy. If a figure has moved, if the deployment has changed, or if you have a higher-quality source, tell us. Every sourced correction is verified before publication.