Google Cloud’s Gemini Live API Launch on Vertex AI Featuring 11Sight

Google Cloud recently announced the general availability of Gemini Live API on Vertex AI, marking a major step forward in real-time, multimodal conversational AI. As part of this launch, Google Cloud highlighted a select group of companies building production-ready AI systems using Gemini Live API - and 11Sight was among them.

In Google Cloud’s official announcement, 11Sight was featured for its work developing voice AI agents that handle real customer conversations at scale, with measurable impact in live environments. The feature included a demo video and quote from Dr. Farokh Eskafi, CTO of 11Sight, outlining how the platform has delivered consistent performance improvements over time.

This recognition reflects the direction 11Sight has taken from the beginning: building conversational AI designed for real operations, not experiments.

What the Gemini Live API Launch Represents

The Gemini Live API launch introduces a new standard for conversational AI in enterprise environments. Built on the Gemini 2.5 Flash Native Audio model and deployed through Vertex AI, the API enables agents that can process and respond to voice interactions with extremely low latency - while maintaining context, natural turn-taking, and conversational flow.

For enterprises, this matters because real customer conversations are unforgiving. Delays, interruptions, or unnatural responses quickly break trust. Systems that perform well in demos often struggle when deployed at scale.

Google Cloud’s launch focused specifically on companies using Gemini Live API in real-world production scenarios, where reliability, performance, and scalability are non-negotiable. 11Sight was featured because its AI agents operate successfully in those conditions every day.

11Sight builds conversational AI agents that manage customer interactions end to end - including booking appointments, qualifying leads, routing calls, and handling follow-ups. These agents operate in environments where conversations are time-sensitive, high-volume, and directly tied to revenue and customer experience.

To support this, 11Sight required an infrastructure that could deliver:

consistently low latency across thousands of simultaneous conversations
natural voice interaction without scripted or robotic behavior
enterprise-grade stability and scalability
the ability to improve performance continuously over time

Gemini Live API on Vertex AI provided the foundation needed to meet those requirements.

As Dr. Farokh Eskafi shared in Google Cloud’s announcement:

“The Gemini 2.5 Flash Native Audio model on Vertex AI gave us the enterprise-grade platform required to rapidly develop our voice AI agents with very low latency. Integrating this solution with our Sentinel AI Agents pushed our call resolution rates from 40% in February to 60% in November.”

This combination of real-time voice AI and continuous performance monitoring is what enabled 11Sight to move beyond proof-of-concept deployments and operate at production scale.

Technology Behind 11Sight’s Voice AI Agents

Gemini Live API and the Gemini 2.5 Flash Native Audio Model

At the core of 11Sight’s voice AI is the Gemini 2.5 Flash Native Audio model. Designed for real-time conversational use cases, the model supports:

fast, low-latency speech processing
natural conversational turn-taking
interruption handling without loss of context
accurate intent recognition from voice signals

This allows 11Sight’s agents to respond fluidly, even in complex, fast-moving conversations.

Vertex AI as the Enterprise Platform

Deploying on Vertex AI ensures that these capabilities are delivered with the reliability and scale required for production workloads. Vertex AI provides:

global infrastructure for consistent performance
enterprise-grade security and compliance controls
scalability across regions and use cases
stability for mission-critical customer interactions

For 11Sight, this meant being able to support customers without compromising on performance as volume and complexity increased.

Why Monitoring and Improvement Matter in Production AI

Building a capable AI agent is only part of the challenge. Maintaining and improving performance over time is what separates production systems from early-stage tools.

Customer conversations change. Call types evolve. New edge cases appear. Without oversight, even well-designed AI systems plateau or degrade.

This is why 11Sight built Sentinel AI Agents as a core part of the platform.

Role of Sentinel AI Agents

Sentinel AI Agents act as a continuous monitoring and quality layer across all 11Sight AI interactions. Rather than relying on static rules or occasional audits, Sentinel operates in real time to:

monitor live conversations
detect breakdowns, stalls, or misrouting
flag recurring failure patterns
feed insights back into system improvements

This approach allows 11Sight to improve performance deterministically - through structured learning and iteration - rather than relying on chance or isolated tuning.

The impact of this approach is measurable.

What This Means for 11Sight Customers

For enterprises and dealerships using 11Sight AI agents, this recognition reinforces several important points:

The AI agents customers rely on are built on proven, enterprise-grade infrastructure
Performance improvements are driven by monitoring and iteration, not static models
The platform is designed to scale without sacrificing reliability or customer experience
The technology stack is aligned with where enterprise AI is headed, not where it has been

In practical terms, this translates to more resolved conversations, fewer dropped interactions, and more consistent customer experiences - even as volume increases.

A Collaborative Partnership with Google Cloud

11Sight’s feature in the Gemini Live API launch reflects a close technical collaboration with the Google Cloud team. The partnership focuses on building and deploying real systems that meet enterprise requirements, rather than showcasing isolated capabilities.

11Sight is grateful for the opportunity to work with the Google Cloud team and to be included alongside other cutting edge companies using Gemini Live API to redefine how conversational AI is deployed in production.

Looking Ahead

The general availability of Gemini Live API on Vertex AI represents an important milestone, but it is not the end of the journey. For 11Sight, it reinforces a long-term commitment to:

building AI agents that perform reliably in real operations
improving outcomes through continuous monitoring and iteration
delivering measurable results rather than experimental features

As conversational AI continues to mature, 11Sight will keep focusing on what matters most: performance, reliability, and real impact for customers.

Go Back