Agents Form Teams. Expertise Gets Embedded. Automation Moves Closer to Action.

🎉 Today marks the 100th edition of Agentic.fyi.

Over the past 100 posts, we've tracked the rise of AI agents, enterprise AI, and physical AI—from experimental demos to production systems. Thank you for reading, sharing, and helping build this community.

Over the first 100 editions, a few posts stood out as favorites:

🔴 Agent Stack Series
Choose Who Gets a Seat at Your Table
A framework for thinking about the agents, tools, and information sources shaping your decisions.

🛠️ 5104 Tinker Lab
Build a Brain Trainer App
A hands-on experiment using AI-assisted development to build a personalized brain-training application.

What happened
IBM and Google announced a cloud partnership focused on building industry-specific AI agents based on Google’s Gemini tools, leveraging IBM’s Consulting Advantage for deployment and hybrid-cloud modernization.

Why it matters
The deal signals that major enterprise players are locking in agentic workflows as a core modernization play, not just an experimental add-on, potentially creating billions in hybrid cloud and agent deployment services.

What’s next
Expect tailored agent solutions across industries (e.g., finance, healthcare, supply chain) and a race among cloud consultancies to embed themselves in mission-critical automation stacks.

What happened
French app publisher MWM and Google Cloud unveiled the “AI Mobile Squad,” a coordinated team of three specialized agents—a Designer, Product Manager and Developer—that turns a user prompt into a production‑ready iOS or Android app in under three minutes. Built on Gemini Enterprise and MWM’s proprietary mobile data, each agent handles discovery interviews, designs mockups and ships native code, sharing context across the squad to streamline workflows.

Why it matters
This marks a shift from single‑agent chatbots to full‑stack agentic teams: instead of just drafting code, the system automates product management, design and development. MWM aims to put a mobile product team within reach of solopreneurs and small businesses, highlighting how agentic AI is moving into mainstream app creation.

What’s next
MWM plans to extend the squad with agents for app‑store optimization, growth, QA and customer support. Rival platforms may respond with their own agent squads as enterprises demand end‑to‑end automation.

What happened
Demis Hassabis, CEO of Google DeepMind, said on a Stanford firesight chat that artificial general intelligence could arrive as soon as 2030 and urged broad societal preparation for its impacts.

Why it matters
A leader from one of AI’s frontier labs publicly tightening AGI timelines reinforces industry and regulatory focus on capability governance and enterprise strategy around future systems.

What’s next
Watch for follow-on statements from other labs and potential shifts in policy and enterprise planning as AGI expectations firm up among practitioners and stakeholders.

What happened
C3 AI announced that Shell is extending its enterprise‑wide predictive maintenance program to include agent‑based root‑cause analysis and remediation. The multi‑year deal builds on Shell’s deployment of C3 AI Reliability across more than 13,000 pieces of equipment, introducing AI agents to diagnose and address anomalies across global operations.

Why it matters
This partnership shows physical AI moving beyond anomaly detection into autonomous problem‑solving. By embedding agents into industrial reliability systems, Shell aims to reduce unplanned downtime and save hundreds of millions of dollars.

What’s next
Watch for similar agentic upgrades across manufacturing and energy as companies seek to automate maintenance. Success could spur regulators to update safety standards for AI‑driven industrial operations.

What happened
The U.S. Army’s C5ISR Center is equipping ground vehicles and robots with a suite of AI‑enabled sensors to detect, classify and report explosive threats during breaching operations. The Ground‑based Multi‑Mission Payload prototype integrates an open AI architecture and supports various platforms—including robot dogs and eight‑wheeled unmanned vehicles—to automate hazard scanning and reduce soldiers’ exposure.

Why it matters
Automating one of the most dangerous battlefield tasks could significantly increase survivability and speed. The modular, open architecture allows the Army to integrate the best algorithms from any source, ensuring adaptability as technologies evolve.

What’s next
Field trials will determine whether the system can reliably replace manual scanning. The Army plans to incorporate new sensors and AI algorithms over time, potentially expanding the concept to other hazardous missions.

AI is moving from isolated assistants to coordinated systems that can build products, run business workflows, support scientific discovery, maintain industrial assets, and operate in high-risk environments. The next competitive advantage will come not from a single model, but from assembling teams of agents, domain expertise, and real-world execution into systems that can reliably deliver outcomes at scale.

Build your own “Agent Squad”. Use Gemini, Claude, or OpenAI Codex and assign each a role: Product Manager, Researcher, Designer, Developer, or QA Reviewer. Give them a simple project—an app idea, marketing campaign, or business process—and compare the results to using a single AI assistant.

Next, introduce domain expertise. Use a specialized model or custom knowledge base to tackle a problem in your field, whether that's healthcare, finance, cybersecurity, sales, or operations. Measure how much performance improves when the agent has access to expert workflows instead of general knowledge.

Finally, identify one repetitive process in your business that could move from assistance to execution. Today's stories suggest the biggest shift is not smarter AI—it's teams of specialized agents working together to deliver real outcomes.

Coming Soon: The First 100 Agentic.fyi Cartoons — a downloadable collection of the cartoons from the first 100 editions.