Google is testing an agent workspace inside Gemini Enterprise
Leaked screenshots show Google testing a dedicated Agent tab inside Gemini Enterprise, pointing to a broader push into multi-step task automation ahead of Google I/O.
Google is apparently testing a new Agent tab inside Gemini Enterprise, according to leaked screenshots published by TestingCatalog. The tab sits directly next to the standard chat interface and introduces a task-focused layout built around goal-setting, connected tools, and multi-step execution.
Everybody is building agentic tools right now, so it only makes sense for Google to do the same.
When a task is started, the interface opens a split view with a chat area and a configuration panel on the right. That panel includes:
- Goal: Define the objective for the agent to work toward.
- Agent: Select which agent handles the task.
- Connected apps: Link external tools and services.
- Files: Give the agent access to relevant documents.
- Require human review: A toggle that controls whether a task executes automatically or waits for manual approval first.
The overall structure is closer to a task execution workspace than a chat product. The “Require human review” toggle lets admins or users decide how much autonomy the agent gets, which matters most when tasks touch live data or external services.
Two related features have also been spotted in testing inside Gemini Enterprise:
- Projects: Persistent, longer-running workflows that carry context over time.
- Skills: Reusable capabilities or custom agent behaviors that can be applied across tasks.
A Gemini desktop app has also been floated as a possibility, though nothing is confirmed. Google is already building a desktop app for AI Studio, and whether the two efforts eventually merge is an open question.
None of this has been confirmed by Google. The screenshots surfaced in Gemini Enterprise accounts, with no indication of a consumer rollout timeline. Google I/O is set for May 19-20, about five weeks away.
Source: Testing Catalog
If you need on-demand GPUs for training, fine-tuning, inference, or running open-source models, give RunPod a try.
- Available hardware: H100, H200, A100, L40S, RTX 4090, RTX 5090, and 30+ more
- Cost: significantly cheaper than AWS or GCP, billed per second, no contracts
- Setup: spins up in under a minute, 30+ regions worldwide
Get the core business tech news delivered straight to your inbox. We track AI, automation, SaaS, and cybersecurity so you don't have to.
Just read what you want, and be done with it.
