• Speeding up agentic workflows with WebSockets in the Responses API — OpenAI breaks down how connection-scoped caching and WebSockets significantly reduce API latency for the Codex agent loop. Better latency fundamentally improves autonomous software engineering. 🔗 Graph: openai, codex, agentic-ai 📅 Published: 2026-04-22 📰 https://openai.com/index/speeding-up-agentic-workflows-with-websockets

• Train separately, merge together: Modular post-training with mixture-of-experts — AI2 demonstrates BAR, a method for post-training language models by independent domain experts that are then merged. This represents a modular upgrade path without risking global model stability. 🔗 Graph: ai2, higher-ed-ai, model-agnosticism 📅 Published: 2026-04-20 📰 https://allenai.org/blog/bar

• Anthropic Launches Opus 4.7 AI Model, Focusing on Coding, Visual Tasks, and Cybersecurity Guardrails — Anthropic’s flagship Claude capability receives an update pushing it further regarding autonomous execution, multi-step work, and coding tasks. 🔗 Graph: anthropic, claude, agentic-ai 📅 Published: 2026-04-21 📰 https://campustechnology.com/articles/2026/04/21/anthropic-launches-opus-4-7-ai-model-focusing-on-coding-visual-tasks-and-cybersecurity-guardrails.aspx

• We're launching two specialized TPUs for the agentic era. — Google Cloud unveils its 8th generation of TPU hardware. These new processors target the unique hardware economics behind executing and serving long-running, process-heavy agentic systems. 🔗 Graph: google, google-cloud-ai, agentic-ai 📅 Published: 2026-04-22 📰 https://blog.google/innovation-and-ai/infrastructure-and-cloud/google-cloud/tpus-8t-8i-cloud-next/

• Introducing workspace agents in ChatGPT — OpenAI brings its Codex-powered workspace agents into ChatGPT’s enterprise offerings, targeting automated complex workflows that scale securely across standard enterprise tools. 🔗 Graph: openai, codex, agentic-ai 📅 Published: 2026-04-22 📰 https://openai.com/index/introducing-workspace-agents-in-chatgpt

• Model Context Protocol (MCP) on Databricks | Databricks on AWS — Databricks introduces MCP to its ecosystem, positioning its AI Gateway as the enterprise control plane for governing MCP servers alongside standard LLM endpoints. 🔗 Graph: databricks, model-context-protocol, ai-governance 📅 Published: 2026-04-22 📰 https://docs.databricks.com/aws/en/generative-ai/mcp/

💡 Signal: This week's core narrative revolves entirely around agentic execution and workflow automation. OpenAI, Anthropic, Databricks, and Google are all actively laying down infrastructure (WebSockets, specialized chips, enterprise routing/governance) specifically designed for agents navigating enterprise IT tooling—which is exactly the trajectory Mission Control and the Enterprise Data Agent are running on.