OpenAI launches Workspace Agents

OpenAI launched Workspace Agents in ChatGPT on April 22, adding shared, cloud-run agents that teams can use for multi-step work inside ChatGPT and Slack. (openai.com) OpenAI said the new agents are powered by Codex and can handle tasks such as preparing reports, writing code, and responding to messages while users are offline. (openai.com) The company described Workspace Agents as an evolution of GPTs, with agents built once, shared across an organization, and improved over time by teams. (openai.com) OpenAI said the product is available in research preview for ChatGPT Business, Enterprise, Edu, and Teachers plans, with role-based controls for who can create, share, and connect agents to tools. (openai.com) (the-decoder.com) The product pushes ChatGPT deeper into workplace software, where companies want agents that can carry context across files, apps, and approvals instead of answering one prompt at a time. (openai.com) (the-decoder.com) That market got more crowded on April 28, when Amazon Web Services said Amazon Bedrock would offer OpenAI models, Codex, and Managed Agents powered by OpenAI in limited preview. (aws.amazon.com) (aboutamazon.com) Amazon said Bedrock customers can access OpenAI models through the same Bedrock services they already use for model access, fine-tuning, and orchestration, rather than moving workloads into a separate stack. (aws.amazon.com) (openai.com) ASAPP added another entry on April 27, saying it launched multiple purpose-built agents inside its customer experience platform to run customer service workflows end to end. (businessinsider.com) (cmswire.com) ASAPP said the release includes agents for building, running, optimizing, and analyzing customer-service operations, aimed at contact centers that want automation across the full workflow. (businessinsider.com) (knowledgenile.com) Across all three launches, the common pitch is managed agents with enterprise controls: shared permissions, tool access, billing, and monitoring handled at the organization level. (openai.com) (aws.amazon.com) (businessinsider.com) The next test is whether companies trust these systems with real workflows, because each new agent layer also adds more tools, permissions, and outputs for information-security teams to audit. (openai.com) (aws.amazon.com)