Band raises $17M to build interaction mesh for enterprise AI

Band, a startup based in Tel Aviv and San Francisco, has exited stealth after raising a $17 million seed round. The company was founded by CEO Arick Goomanovsky and CTO Vlad Luzin and says it will use the funding to build an interaction mesh that governs autonomous AI agents operating across enterprise networks.

The interaction mesh is a software layer that sits between autonomous actors and corporate systems. Band intends for the layer to manage coordination, permissions and data flows, and to enforce routing, error recovery, authority boundaries and audit logging while the agents perform tasks.

Band describes three factors driving demand for the mesh. Autonomous agents are running in production to handle tasks in engineering, support and security. Enterprise environments are heterogeneous: teams use different frameworks, models run on multiple cloud platforms, and communication protocols vary under separate ownership. Standards work for model handshakes, such as the Model Context Protocol and early A2A efforts, defines how models exchange data but does not manage production routing, governance or runtime safety.

The company highlights financial and operational risks when agent interactions are unmanaged. Point-to-point integrations maintained by developers can increase long-term maintenance. Uncontrolled multi-agent exchanges can trigger repeated model calls and high compute or token costs; Band plans to include hard circuit breakers that terminate interactions exceeding set token or compute thresholds.

Band also points to risks to data integrity and compliance when autonomous systems interact with legacy corporate architectures. Financial firms and healthcare providers often use on-premises data warehouses, mainframes and custom ERP systems. Concurrent automated actions by different agents can create database locks or conflicting records. Vector databases used for retrieval-augmented generation are often isolated for specific use cases; passing context between those stores without cryptographic verification can reduce data fidelity and obscure data lineage.

To address these issues, Band says its mesh will act as a security perimeter where access controls and delegation chains can be inspected and enforced at the communication layer. The platform will log interactions with cryptographic proofs so auditors can trace automated decisions to their origin. Band designs the mesh to operate across frameworks and clouds so teams can keep specialized models while sharing a governed runtime.

Band says the $17 million seed will fund product development and initial enterprise integrations. The company is targeting organizations that run autonomous agents across fragmented cloud and data stacks and aims to give data officers a place to enforce access controls and provenance rules at the interaction layer.