Agents Connect. Systems Coordinate. Control Scales.

What happened
ServiceNow and Google Cloud deepened their partnership at Cloud Next ’26, unveiling unified AI agent solutions that span 5G networking, retail and IT operations. Agents built on Gemini Enterprise and the ServiceNow AI Platform can now interoperate via open protocols (Agent‑to‑Agent, Agent‑to‑UI and Model Context Protocol) and share data through BigQuery and ServiceNow’s Workflow Data Fabric. Governance comes via a shared registry that tracks every agent’s actions across both platforms.

Why it matters
Enterprise AI agents often live in silos. By standardizing how agents talk to each other and building a unified control plane, Google and ServiceNow aim to turn autonomous workflows into a managed, cross‑platform workforce. This could reduce downtime in networks and retail operations by letting agents detect anomalies, triage fixes and dispatch technicians without human hand‑offs.

What’s next
Preview solutions for autonomous network and retail operations are available now, with broader releases to follow. Expect other vendors to adopt open protocols like MCP and unify agent governance as enterprises demand cross‑platform interoperability.

What happened
Data‑security company Rubrik announced two integrations at Cloud Next: a Cloud SQL cyber‑resilience module and Rubrik Agent Cloud for the Gemini Enterprise Agent Platform. The SQL integration provides immutable backups and cross‑region recovery for managed PostgreSQL, with policies to enforce compliance. Rubrik Agent Cloud adds a governance layer for AI agents, including a semantic AI governance engine, auto‑discovery of agents, risk and permissions visibility and an “Agent Rewind” feature to roll back destructive actions.

Why it matters
As enterprises deploy more autonomous agents, security and compliance become bottlenecks. Rubrik’s control layer promises to monitor agent behavior and enforce policies in real time, addressing concerns that 90 % of IT leaders faced cyberattacks last year.

What’s next
Gartner predicts 40 % of enterprise applications will integrate task‑specific agents by 2026. Partnerships like this, combining agent governance with data‑resilience, will likely proliferate as firms seek both speed and safety.

What happened
Google introduced the “Agentic Data Cloud,” a platform designed to act as a dynamic reasoning engine for AI agents. It combines a universal context engine to prevent hallucinations, a Data Agents Kit that brings agentic skills into developer tools like VS Code, and a cross‑cloud Lakehouse that makes data from other clouds accessible without migration. Specialized agents for data engineering, data science and database observability accompany the launch.

Why it matters
Current data platforms are passive repositories. Google’s approach turns them into systems of action, enabling agents to understand company‑specific metrics and execute data‑engineering tasks autonomously. The universal context engine and Model Context Protocol aim to ground agents in governed data and reduce errors.

What’s next
Google says the Agentic Data Cloud will underpin its own Deep Research Agent. If successful, expect other cloud providers to offer similar agent‑centric data stacks and for developers to adopt the Data Agents Kit to build agentic workflows into existing codebases.

What happened
OpenAI announced “Workspace Agents” — Codex‑powered agents for ChatGPT Business and Enterprise plans that can run long‑lived workflows across third‑party apps. Teams can design or choose agent templates to handle tasks in Slack, Google Drive, Salesforce, Notion and other tools, and agents can draft emails, pull data, prepare reports and make presentations. Because they run in the cloud on Codex, they can write and execute code, persist memory, schedule recurring tasks and continue work even when the requester is offline.

Why it matters
Previous GPT‑based assistants were session‑bound and required “babysitting.” Workspace Agents shift AI from a chat aide to a collaborative worker that persists across projects and channels. This could automate routine processes in enterprises and turn agent hand‑offs into reusable, shared workflows.

What’s next
Workspace Agents are free until May 6 and will adopt credit‑based pricing afterward. OpenAI plans to add triggers, dashboards and deeper integrations; expect other model providers to answer with their own persistent agent offerings.

What happened
SpaceX revealed it could acquire Cursor, a leading AI coding assistant, for a staggering $60 billion later this year. The deal would mark one of the largest AI acquisitions to date.

Why it matters
AI-powered software development tools are now strategic assets for tech giants. SpaceX’s move highlights the growing importance of AI in automating and accelerating code, with ripple effects across the entire software industry.

What’s next
Expect more mega-deals as companies race to secure the best AI talent and tools. Cursor’s fate could reshape the competitive landscape for developer-focused AI.

What happened
Sony’s AI research division unveiled Ace, an autonomous robot that competes with and sometimes defeats top‑level human table‑tennis players. Ace uses nine synchronized cameras, multiple vision systems and a learning‑based control algorithm to read ball spin and react faster than human reflexes. It has already won three out of five matches against elite players and has beaten professionals in subsequent exhibitions.

Why it matters
AI has long surpassed humans in digital games, but physical sports require real‑time perception and motor control. Ace demonstrates that robots can achieve expert‑level performance in dynamic, adversarial environments. The techniques could transfer to manufacturing, service robotics and safety‑critical domains.

What’s next
Researchers are refining Ace to better adapt to human strategies. As physical AI continues to push into sports, expect similar systems to emerge in industry and healthcare, where fast, precise interactions are essential.

What happened
Chef Robotics announced a physical AI application that automates produce packing, using piece‑picking and scooping capabilities to place fruits like oranges and pears into clamshells and portion scoopable produce such as corn and peas into meal trays. The system employs AI‑powered computer vision to assess each item’s position, shape and orientation and uses tray‑tracking to ensure consistent placement. It can center items, arrange multiple pieces in one pass and stack layers for deep trays.

Why it matters
Produce packing has resisted automation because fruits and vegetables vary in size and texture. Chef’s application reduces labor, increases throughput and delivers retail‑ready presentation. It shows how physical AI models trained across diverse environments can adapt to real‑world variability without pre‑sorting.

What’s next
The produce packing capability is available in the U.S., Canada and the UK under Chef’s robotics‑as‑a‑service pricing model. Expect similar physical AI solutions for meatpacking and other food‑handling tasks, and more food manufacturers adopting robotic systems to address labor shortages.

Agents are no longer isolated tools—they’re becoming interconnected systems that act across platforms and persist over time. The real shift is from building smarter agents to managing coordinated, governed networks of them.

Pick a workflow that spans multiple tools (code → data → action).

Use Cursor to generate or modify code for a real task (script, API call, or automation)
Use ChatGPT or Claude to plan the workflow and define steps
Imagine it pulling data from BigQuery and triggering actions in ServiceNow

Then ask: what permissions, monitoring, and rollback would I need if this ran continuously across systems?

You’ll see the shift: it’s not one agent doing a task—it’s a coordinated system executing work.