PubNub has introduced Blocks.ai, a global network designed to connect and control AI agents across diverse frameworks, providers, and APIs. This infrastructure allows developers to reach existing AI agents regardless of their hosting location, addressing a significant barrier in enterprise adoption. The Blocks Network supports all AI agent use cases without requiring inbound ports, tunnels, DNS changes, or firewall modifications, providing a pathway for agents to move from isolated local environments into production.
Connecting AI Agents to Specialist Capabilities
Blocks.ai functions as a global agent control plane, enabling agents to communicate and share efficiencies without requiring developers to rebuild integrations. Developers who do not host their own agents can utilize the network to discover capability-specific agents via open-source SDKs. Instead of managing unique API keys for every provider, applications can call AI functionality using standard function calls.
This connectivity extends beyond simple access; it allows a builder’s agent to call other agents on the Blocks Network. Once connected, an agent can access discoverable capabilities such as code review, transcription, PDF and QR generation, data extraction, and streaming compute. This capability allows agents to interact with specialized functions without the builder maintaining separate integrations for each service.
Control, Privacy, and the Blocks Network Infrastructure
The Blocks Network is built upon PubNub’s real-time backbone, incorporating zero-trust security, SOC 2, and GDPR compliance. This architecture ensures that the agent remains private and under the builder’s control, even when it is reachable from external frontends.
Todd Greene, CEO of PubNub, stated that the network allows agents to be connected once and available to any frontend or user without rebuilding the underlying infrastructure. The platform supports the agent staying on the builder's machine while remaining globally accessible. Open-source SDKs for TypeScript and Python are available under the Apache-2.0 license, ensuring the code is auditable and easy to integrate into existing enterprise stacks.
Operational Relevance for Enterprise AI Deployment
The introduction of Blocks.ai directly addresses the challenge of agent isolation, a factor cited by McKinsey as contributing to fewer than 10% of enterprise agents reaching production. By providing a secure, network-level control plane, PubNub aims to remove the brittle infrastructure work typically required to make locally hosted agents reachable from custom frontends.
The platform is designed to be accessible immediately. Blocks.ai is available to connect today, allowing users to make AI agents globally callable in seconds using the open-source SDKs or the CLI. This deployment model allows businesses to leverage AI capabilities without the complexity of managing inbound network configurations for every agent deployment.
Key Takeaways
- Blocks.ai functions as a global network to connect and control agents across all agent frameworks, providers, and APIs.
- The Blocks Network supports AI agent use cases without requiring inbound ports, tunnels, DNS changes, or firewall modifications.
- The platform offers open-source SDKs for TypeScript and Python under the Apache-2.0 license, maintaining agent control and data privacy.
TechInsyte's Take
In our view, the launch of Blocks.ai signals a critical shift in the enterprise AI adoption lifecycle. The primary bottleneck for many organizations is not the model itself, but the infrastructure required to deploy and connect it securely. By abstracting the network layer—removing the need for complex firewall rules and port management—PubNub is positioning itself as the connective tissue for the "Internet of Agents." This move suggests that future enterprise AI integration will rely less on monolithic vendor stacks and more on a decentralized, auditable network where specialized capabilities can be discovered and called as standard functions. For CIOs and CTOs, this offers a path to productionizing AI agents without the immediate risk of deep infrastructure overhaul.
Source: Businesswire
