Ai4 2026
What's happening at our booth
Live inference-stack demo. STT, LLM, and TTS running on one private network we own. Sub-200ms round-trip latency, frontier open-weight models (GLM-5.2, Kimi K2.6, MiniMax-M3) available day zero. Swap models mid-demo with our LLM + STT/TTS Router, without rewriting a line of code.
Real-time AI voice agent calls handled end-to-end on Telnyx infrastructure. Agents that pick up because every layer underneath them is ours: compute, network, and carrier-grade telephony.
5-minute stack-consolidation modeled with one of our reps on your specific call and token volume. Bring your current vendor list.
Live Panel Session
Wednesday, August 5 · 10:30–11:15 AM PDT. Panel: "Engineering the Infrastructure Stack for Enterprise-Scale Agents" (Track: Infrastructure: AI Agents).
Featuring David Casem (CEO & Co-Founder, Telnyx) alongside Lu Zhang (Fusion Fund), Ilya Kirnos (SignalFire), Mark McNeill (OneSource Cloud), and Hok Hei Tam (Montai Therapeutics).
What it actually takes to run agent infrastructure at scale: owned vs. rented, latency, and eliminating multi-vendor sprawl.
Come talk to us about
Shipping enterprise Voice AI in production without stitching together five vendors. Inference, voice, and telephony on one platform.
The flexibility to swap frontier models anytime. Leading open-weight models hosted natively on infrastructure we own, no rewrites, no lock-in.
Fast inference, not just local inference. Dedicated GPUs in the US, EU, and APAC, so your requests don't queue behind someone else's workload.
Structural cost savings. Frontier inference starting at $0.21/1M tokens, with no cloud markup baked into every token.
Consolidating your agent stack to one platform. Compute, inference, voice, and messaging, one API, one DPA, one bill.
📍 Booth #1332 · The Venetian Expo, Hall BC (2nd Floor) · August 4–6, 2026
See you there!
Want a dedicated walkthrough? Request to join with your contact information and a member of our team will reach out to schedule a private demo during Ai4 (we're capping these at 20 slots). Bring your stack diagram. We'll walk through the layers you can remove.