2026 is the year of agentic AI adoption. Imagine a world where intelligent agents handle many tasks, freeing us to focus on what truly matters. The pace of innovation we see today makes us question how to learn and adapt in this new AI era, bringing excitement to some and fear of the future to others.
One of the top AI leaders in the industry, John Capabianco, shared a perspective on the current state of AI that stuck with me for days:
The best way to get on the train is when it’s parked at the station. This train is no longer parked at the station. It's moving and moving quietly rapidly, so now it’s up to you to chase down this train and try to get on it.
Back to the past
Can you imagine a time when devices couldn’t communicate easily due to proprietary protocols? Look at where we are now: a Cisco router can connect to a Huawei switch, which in turn can connect to a Mikrotik wireless access point. Thanks to the OSI model and widespread standardization, devices can communicate seamlessly today. Standardization hasn’t stifled innovation; it has accelerated it. Every vendor, regardless of competition, can carve out their slice of the market. That’s the power of standards, and it’s shaping the industry we see today.
The Agentic AI era is an Internet of Agents (IOA)
Imagine an internet where agents can discover each other and exchange information securely. This is exactly what ANGTCY is doing. When we talk about agents, we often forget the builders laying the foundations for the future of internet agents.
ANGTCY is doing for the Internet of Agents what standardization did for networking. Initially developed by Outshift, with support from Cisco and other industry leaders like LangChain and Galileo as maintainers, what began as a simple GitHub repository has grown into a project shaping the future of the Internet of Agents. Today, more than 75 companies have joined the effort.
Avoiding silos between agents makes the future of agentic AI open, inclusive, and fertile for greater innovation.
The AGNTCY framework
- Agent discovery: AGNTCY leverages the open agent schema framework (OASF) allowing any agent to discover and understand capabilities of others, think of the Cisco Discovery protocol developed in 1994, it’s helped Directly Cisco Devices to discover each other by sharing like device IDs, platform types, interface details, and IP addresses to simplify troubleshooting and network mapping it was proprietary then LLDP came to solve the vendor-neutral problem, Think now of the OASF like LLDP even if they have different purpose today for your understanding.
- Agent Identity: Enables agents to operate securely across organizations through cryptographically verifiable identities and controlled access.
- Agent Messaging: Enables multi-modal, human-in-the-loop, and quantum-safe communication through SLIM (Secure Low-latency Interactive Messaging).
- Agent observability: Offers comprehensive observability tools to monitor, assess, and debug complex multi-agent workflows across different vendors and frameworks.
- Protocol Integration: Compatible with all agent protocols, including Agent-to-Agent (A2A) and Model Context Protocol (MCP). Both A2A agents and MCP servers can be discovered via AGNTCY directories and monitored using AGNTCY observability tools.
AGNTCY VISION: Agentic AI will supercharge human work. Enterprises must build agentic workflows and applications that integrate both internal and third-party agents to harness AI’s full potential, speed up operations, and unlock substantial productivity gains.
Top Business Solutions & Impacted Companies
| Solution Category | Key Problem Solved | Core Technology Used | Formative & Participating Companies |
| Agent Interoperability | Prevents vendor lock-in; allows agents from different frameworks | A2A (Agent-to-Agent) & MCP (Model Context Protocol) | Cisco, Google Cloud, Red Hat, LangChain, Galileo |
| Identity and Trust | Ensures that autonomous agents have verifiable "passports" to access secure corporate data and tools. | AGNTCY Identity (Cryptographically verifiable IDs) | Dell Technologies, Oracle, Cisco |
| Discovery and routing | Creates a "DNS for Agents" so a system can find the best specialized agent for a specific task | OASF (Open Agentic Schema Framework) | Cisco, Dell, Google Cloud |
| Secure messaging | Provides low-latency, quantum-safe communication between agents, humans, and tools | SLIM (Secure Low-latency Interactive Messaging) | Cisco, Red Hat, Dell |
| End-to-End Observability | Tracks the "reasoning" steps of multiple agents to debug errors and ensure reliability in production. | AGNTCY Observe SDK | Galileo, Cisco, Google Cloud |
AGNTCY was originally open-sourced by Cisco in March 2025 and officially donated to the Linux Foundation in July 2025.
Agentic AI hype or business opportunity?
While we live between skepticism and excitement in this AI era, agentic AI is steadily taking shape. The real divide will soon be between agentic and non-agentic companies.
Organizations that adopt agentic AI, paired with experienced teams and strong human judgment, will stand out and lead the future of business.
As of early 2026, the market is valued at roughly $6–7.6 billion and is expected to grow significantly, reaching an estimated $47 billion to $136 billion between 2030 and 2035.
Real-World Production Use Cases for AGNTCY
AGNTCY lays the foundation for agentic AI, and several major technology companies are already adopting it.
- SRE and platform engineering: The Cisco Outshift team uses a multi-agent system to manage infrastructure, automating 30% of SRE workflows, including incident response and ticket coordination across systems like ServiceNow and Salesforce.
- CoffeeAGNTCY Reference Architecture: Digital Realty uses the AGNTCY project to validate and demonstrate how multi-agent systems (MAS) can scale across distributed, hybrid-cloud environments.
- Network configuration: Telecom providers, including Swisscom, are using AGNTCY in proprietary tools to automate and validate network migrations and architectural changes.
Why Agentic AI matters for businesses
Agentic AI allows enterprises to delegate routine, repetitive, or cross-system tasks to autonomous agents. This not only accelerates workflows but also reduces human error, frees talent for higher-value work, and enables scalable, adaptive operations.
In an AI-driven economy, businesses that fail to integrate agentic workflows risk falling behind competitors who are leveraging agents to innovate faster, respond to market changes instantly, and create new revenue streams.
Conclusion
The era of agentic AI is no longer approaching; it’s underway. By embracing open standards and collaboration, initiatives like AGNTCY are ensuring that the Internet of Agents remains interoperable, inclusive, and fertile for innovation.
The choice is yours: chase the train and shape the future, or watch it speed by.