BLACKBOX chooses ElevenLabs Agents over OpenAI to power Logger and Robocoder

BLACKBOX AI is pioneering a new category of enterprise software - fully autonomous, voice-first agents that manage everything from coding to infrastructure without human bottlenecks. Today, they power over 30 million developers and their two primary voice enabled products are: 

• Logger: agent that continuously monitors logs and escalates issues via phone call with a conversational agent and proposes or executes fixes.
• Robocoder: voice-driven coding agent that explains, writes, and executes tasks.

Challenge

As a leader in deploying fully autonomous coding agents at global scale, BLACKBOX AI faced a unique challenge: how to make those agents feel truly human.

Before adopting ElevenLabs, BLACKBOX AI experimented with multiple AI providers, including OpenAI, to build conversational interfaces for their agents. However, they encountered three key limitations: 

• Unnatural conversations, leading to lower user engagement.
• A limited set of only 10 voices, restricting multilingual and locale-specific use cases.
• High costs at enterprise scale

Why BLACKBOX chose ElevenLabs over OpenAI

After A/B testing the two providers, ElevenLabs stood out as the only one that could match BLACKBOX AI’s standard for natural, contextual, low-latency voice interactions. In particular, BLACKBOX AI chose ElevenLabs for three main reasons:

• Superior conversational quality: Users consistently preferred ElevenLabs voices in A/B testing, with higher adoption and engagement. Natural turn-taking and more realistic delivery made conversations flow more smoothly than with other providers.
• Global voice coverage: ElevenLabs supports a significantly wider selection of voices - more than 5,000 compared to OpenAI’s 10 - across different languages and accents. This breadth ensures more accurate, natural interactions globally.
• Scalable cost-effectiveness: ElevenLabs offers a pricing model that makes large-scale deployment feasible, an important factor for a company like BLACKBOX AI that powers over 30 million users.

We A/B tested OpenAI’s and ElevenLabs’ voice agent offering and found that usage was better with ElevenLabs. Conversations felt more natural and the voice selection across different languages mattered.

– Robert Rizk, Co-Founder & CEO at BLACKBOX AI

Implementation

BLACKBOX integrated ElevenLabs Agents across two products:

Logger

• 24/7 monitoring of server logs.
• Autonomous phone calls to on-call engineers when incidents occur.
• Conversational troubleshooting with clear context and proposed fixes.
• Authorized remediation to reduce downtime.

Robocoder

• Voice-driven ai coding agent across IDE, WhatsApp, and phone.
• Real-time code explanations for faster onboarding and knowledge transfer.
• Creation of AI voice apps, enabled by ElevenLabs APIs.

Results

Voice-first workflows are now a core part of BLACKBOX AI’s product experience:

• Engineers resolve incidents faster with real-time conversational alerts
• New devs onboard quicker via voice-guided codebase explanations
• Usage, retention, and developer satisfaction have all increased

Looking ahead

BLACKBOX AI sees voice-first agents as the next wave of enterprise software. Logger and Robocoder are only the beginning - the company plans to expand into more multilingual, real-time, and large-scale deployments powered by ElevenLabs, in line with its mission to remove human bottlenecks and accelerate innovation in software development.