VioX AI BlogFrontier AI dispatches from VioX.https://www.viox.ai/blog2026-04-29T00:00:00ZVioX AIhttps://www.viox.aiMistral Medium 3.5 ships remote agents — here's why we still wouldn't route production traffic through themhttps://www.viox.ai/blog/mistral-medium-35-remote-agents2026-04-29T00:00:00Z2026-04-29T00:00:00ZMistral just shipped Medium 3.5 with remote agents that execute code server-side. We explain why VioX still orchestrates tool calls client-side — and when remote execution makes sense.OpenAI on AWS Bedrock — one day after the Microsoft splithttps://www.viox.ai/blog/openai-on-aws-bedrock-one-day-after-microsoft-split2026-04-29T00:00:00Z2026-04-29T00:00:00ZMicrosoft and OpenAI dissolve exclusivity. 24 hours later, AWS announces OpenAI models on Bedrock plus a jointly-built agent service. The math: never run on one vendor.The 6-layer agentic stack we deploy for SMBshttps://www.viox.ai/blog/six-layer-agentic-stack2026-04-29T00:00:00Z2026-04-29T00:00:00ZReasoning at the bottom, business outcomes at the top. The architecture we run inside every VioX OS deployment — and why each layer earns its place.Evals on day zero — why VioX OS ships every agent with a test suitehttps://www.viox.ai/blog/evals-on-day-zero2026-04-28T00:00:00Z2026-04-28T00:00:00ZAn agent without evals is a complaint waiting to happen. Here's the discipline we hard-code into every VioX OS deployment, and the eval suite every one of our eight agents ships with.We migrated Goldie from Retell to ElevenLabs in 4 days. Here's the diff.https://www.viox.ai/blog/migrating-goldie-retell-to-elevenlabs2026-04-27T00:00:00Z2026-04-27T00:00:00ZThe catering concierge agent for Golden Plate was running on Retell. We ported it to ElevenLabs ConvAI in four working days. What changed, what broke, what got better — and the playbook we'll use next time.