How Vivint Uses Agentforce to Autonomously Resolve 20% of Customer Issues

Vivint is the only fully vertically integrated smart home company in the U.S., handling the design, manufacturing, sales, installation, service, and monitoring of its own devices for more than 2 million customers. Its platform combines security cameras, smart locks, sensors, and thermostats into a single app, with 24/7 monitoring and support. As part of NRG, Vivint had grown into a company where customer service scale was both a competitive differentiator and an operational cost challenge: with 80% of service interactions handled by phone, rising customer demand threatened to outpace support capacity.

The challenge was fundamental to the smart home category. With an average of 14 devices per customer, support interactions ranged from simple payment questions to complex multi-device troubleshooting. High-volume, repetitive requests—billing inquiries, shipment tracking, appointment scheduling, common device issues—were consuming agent time that would be better spent on complex cases requiring human judgment. The smart home market was projected to grow more than 20% annually through 2030, which meant the problem would only compound without a structural solution.

Vivint worked with Salesforce partner NeuraFlash to design and deploy AVA, an Agentforce-powered AI agent that went live to select customers in June 2025. AVA reasons across a knowledge base of more than 1,000 articles, customer and device data from Agentforce Service, and real-time IoT telemetry from Vivint’s proprietary systems via a custom API. Using existing Salesforce Flow automations, AVA handles four categories of interactions autonomously: generating secure payment links for billing, following manufacturer-specific troubleshooting procedures using live device data, pulling real-time shipment tracking information, and integrating with Vivint’s IFS scheduling system to reschedule or cancel technician visits. When an issue exceeds AVA’s scope, it routes the customer to a live representative with full conversation history and a generated briefing summary.

AVA achieved a 20% autonomous resolution rate and drove 28% growth in chat volume—making chat one of Vivint’s highest-performing CSAT channels. Salesforce forward-deployed engineers embedded with Vivint’s team during deployment, iterating on AVA’s reasoning behavior through session traces and conversation monitoring. “Salesforce’s forward-deployed engineers have been critical—working side by side with us, understanding how fast the technology is changing, and helping us tackle use cases for our customers and employees,” said Ryan Gee, SVP of Engineering.

Vivint is now extending Agentforce Voice to bring AVA’s full capability set—troubleshooting, payments, shipments, appointments—to phone calls, the channel that handles 80% of interactions. Because every Flow automation, Apex integration, knowledge article, and API connection built for chat transferred directly to voice, deployment took weeks rather than months. Vivint is also exploring proactive AI service: using IoT telemetry combined with CRM data to identify and resolve issues automatically before customers notice them, shifting the model from reactive support to continuous home intelligence.