🫗 Spilling the Sauce: Voice Agents 🫗
Just published a deep-dive Substack on building production-ready, compliance-safe AI voice agents — no pricey black-box APIs required.

🔗 buff.ly/zSRIgEW

If you’re building voice CX or just tired of vendor lock-in, this one’s for you.

Mistral AI recently released Voxtral, a series of state‑of‑the‑art speech understanding models. I'll be diving into these over the next few weeks in my substack and comparing them to existing open source options for TTS/LLM/STT systems. Subscribe here: buff.ly/aCkGU9O

This whole “don’t call it prompt engineering! It’s context engineering now!” talk seems pointless. These are completely different domains within an AI system.

AI engineering != context engineering != prompt engineering

Just like a square can be a rectangle but a rectangle can't be a square.

I tried 3 new (to me) open‑source tools in my voice AI stack so you don’t have to:
jambonz (jambonz.org)
@livekit.io
vocode (vocode.dev)