Former Vercel executive says voice agents will be won on infrastructure, not models [Sponsored]

2026-05-18 · Source: Tech.eu - Tech.eu · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Software Development & Engineering · Depth: Intermediate, medium

Summary

SLNG, a UK voice AI startup, is challenging the prevailing pricing model in the voice agent industry, which its CEO, Luke Miller, describes as "token-maxing" and systematically overcharging enterprises. Having raised €3.3M in pre-seed funding, SLNG is developing an execution layer, dubbed "the Vercel for voice agents," that sits between existing voice agent orchestrators (like LiveKit or Pipecat) and underlying AI models. This platform intelligently routes speech-to-text, text-to-speech, and LLMs in real time, selecting optimal models, handling in-region routing, failover, and compliance across 11 sovereign regions. SLNG claims its approach reduces model costs by over 50%, cuts latency by more than half, and improves target outcomes by avoiding unnecessary calls to expensive frontier models, particularly benefiting sectors like financial services and healthcare with strict data sovereignty requirements.

Key takeaway

For CTOs and VPs of Engineering building voice agents, your focus should shift from solely powerful models to an optimized execution layer. Implementing a platform like SLNG can significantly cut operational costs and latency while improving customer outcomes by intelligently routing calls and ensuring data sovereignty, rather than over-relying on expensive frontier LLMs for every interaction.

Key insights

Voice agent success hinges on optimized infrastructure and intelligent model routing, not just raw model power.

Principles

• Optimize compute, don't maximize it.
• Consistency and reliability outweigh raw model power.
• Data sovereignty drives enterprise AI adoption.

Method

SLNG's execution layer coordinates real-time speech-to-text, text-to-speech, and LLMs, intelligently routing calls to appropriate models and handling failover and compliance across sovereign regions.

In practice

• Integrate an execution layer for voice agent cost reduction.
• Prioritize in-region data processing for compliance.
• Utilize lighter models for deterministic responses.

Topics

• Voice AI
• SLNG Platform
• Execution Layer
• Data Sovereignty
• Cost Optimization

Best for: CTO, Entrepreneur, VP of Engineering/Data, AI Engineer, MLOps Engineer, AI Architect

Related on AIssential

• 8 billion tokens a day forced AT&T to rethink AI orchestration — and cut costs by 90% — Orchestrating specialized small language models with LLM "super agents" dramatically cuts costs and boosts performance.
• Who Owns the AI: Why Vanity Deployments Are Enterprise’s Most Expensive Mistake — Prioritizing AI optics over robust data infrastructure and compliance leads to expensive enterprise AI failures.
• Coval raises $28M as enterprises push voice agents into production — Voice AI agents demand specialized, scalable testing and monitoring infrastructure to manage unique failure modes and ensure reliable production deployment.
• 🔴 LIVE: Google’s $15B AI Hub Faces Protests | AI Chip Crisis & G7 AI Battle | Front Page — The global AI boom is creating significant infrastructure, supply chain, and governance challenges, while also driving innovation in specialized areas like voice AI.
• Architecting Modern AI Systems: Platforms, Agents, and Integration — Open-source models and sovereign AI platforms offer cost control, data residency, and output flexibility over closed-source APIs.
• Building Scalable AI Voice Agents: Architectures, Latency & Best Practices — Voice agents are fundamentally a latency problem disguised as an AI problem, where speed and natural conversation outweigh raw model intelligence.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Tech.eu - Tech.eu.