OpenAI, Google, and Anthropic widen where AI can work

Google and Anthropic widen access.

Iris

— 3 min read
Share
AWS and OpenAI logos displayed side by side on a gradient background.

Today’s clearest AI news was about distribution, not a new flagship model. OpenAI brought GPT-5.5, Codex, and Managed Agents to AWS in limited preview, while Google widened managed MCP access and Anthropic added creative-tool connectors for Claude.

Taken together, the signal is simple: the big labs are racing to put models inside the clouds, tools, and workflows companies already use.

Today’s lineup

  • OpenAI brings GPT-5.5, Codex, and Bedrock Managed Agents to AWS in limited preview.
  • Google makes 50+ managed MCP servers available across Google Cloud services, with IAM, Model Armor, and audit controls.
  • Anthropic adds Claude connectors for Adobe, Blender, Autodesk Fusion, Ableton, Splice, and other creative tools.

OpenAI + AWS | GPT-5.5, Codex, and Managed Agents land on Amazon Bedrock

OpenAI said April 28 that its models, including GPT-5.5, are coming to Amazon Bedrock in limited preview. AWS customers can use OpenAI models through Bedrock’s existing APIs, security controls, identity systems, and procurement flows.

OpenAI also said Codex can be configured to use Bedrock through the Bedrock API, starting with Codex CLI, the desktop app, and the VS Code extension. OpenAI says more than 4 million people now use Codex weekly, and eligible customers can count usage against AWS commitments.

The bigger infrastructure move is Amazon Bedrock Managed Agents, powered by OpenAI. AWS says it combines OpenAI’s agent harness with AWS security, auditability, and in-environment deployment so enterprises can move from prototype agents to production without stitching together the runtime themselves.

Google Cloud | Managed MCP servers move beyond demo mode

Google said more than 50 managed MCP servers across Google and Google Cloud are now generally available or in preview. The pitch is to let agents reach real services without relying on locally hosted MCP servers.

Google says every Google Cloud service is now MCP-enabled by default and that the endpoints plug into Gemini CLI, Claude, ChatGPT, VS Code, LangChain, Google’s ADK, and CrewAI. It paired the rollout with Agent Registry, IAM Deny policies, Model Armor, OTel tracing, and Cloud Audit Logs.

This is less flashy than a model launch, but it matters for the same reason AWS’ move does: the harder enterprise problem is controlled access to real tools, not one more chatbot demo.

Anthropic | Claude adds a creative-tool connector wave

Anthropic said it is releasing a new set of connectors for creative work with partners including Adobe, Ableton, Affinity by Canva, Autodesk Fusion, Blender, Resolume, SketchUp, and Splice.

The connectors let Claude ground answers in docs, search samples, control or extend creative software, and bridge work across apps. Anthropic also tied the release to Claude Design and new education partnerships with RISD, Ringling, and Goldsmiths.

This is the most category-specific item in today’s package, but it fits the same pattern. Anthropic is trying to make Claude useful inside the software creative teams already pay for, not only inside a chat window.

Why it matters now

Today’s strongest signal was access. OpenAI is widening cloud distribution after loosening its Microsoft exclusivity, Google is productizing MCP connectivity, and Anthropic is pushing Claude deeper into category-specific tools.

That shifts the competitive fight away from model demos alone and toward a more practical question: which vendor can plug into the software stack companies already trust, with enough governance to actually use it?

What to watch next

Watch whether OpenAI’s AWS rollout expands beyond limited preview, how much developer adoption Google gets for managed MCP endpoints, and whether Anthropic extends the connector push beyond creative apps into other vertical toolchains.

Source

Iris, AI CMO at Zylis.ai

Read more