All agents, one network: Why we invested in BAND

Over the last year, one pattern has become impossible to ignore, software is becoming agentic.

Everywhere we look, teams are spinning up agents to plan features, write code, review pull requests, triage incidents, manage tickets, and automate workflows. Our day to day has turned into a distributed workforce of specialized agents.

But for all that progress, there’s still one tiny, tiny problem: none of these agents actually talk to each other.

If you look at a typical engineering team, they copy outputs between tools, re-explain context in every chat, wire up glue code, and hope nothing breaks as workflows scale.

BAND is building the missing layer for this world, a universal communication layer where agents from any framework can talk, share context, and collaborate like a real team  with humans fully in the loop.

That’s why we invested.

Today’s agent stacks all suffer from the same three issues:

1. Silos by design- Each agent typically lives in its own universe with a different framework, runtime, or SaaS product, with its own memory and tools. There is no default way for these agents to discover each other, exchange structured messages, or share state.
2. Developers as the integration layer- When teams want agents to work together, they resort to manual “plumbing”. As the number of agents grows, the system becomes fragile and expensive to maintain.
3. No governance or visibility- Once agents start acting on behalf of users and systems, enterprises need to know who did what, where, and under which policy. Today, that governance is scattered across tools, if it exists at all.

This is especially painful in software development, where teams already rely on multiple coding assistants across planning, implementation, review, and testing. Agents can help at every stage, but they can’t coordinate the work end‑to‑end.

BAND was born directly out of this gap.

BAND brings you a communication layer for AI agents

BAND is building what we see as a new foundational layer in the AI stack, a universal communication and coordination layer for agents. Instead of connecting agents with bespoke integrations, BAND gives them a shared environment where they can:

• Discover each other across frameworks, clouds, and organizations
• Exchange structured messages in real time
• Share context while preserving each agent’s own tools, models, and memory
• Delegate and coordinate work as a connected workforce, not isolated bots

In practical terms, this means your planning agent can hand off work to an implementation agent, the implementation agent can then loop in a review agent, and all of them can operate with shared context and clear rules. Without you hand-coding each connection.

Copilot era OUT workforce era IN

We’re moving from the copilot era (“one assistant per user”) into the workforce era (“many agents per team, per workflow”).

Three shifts make BAND’s timing compelling:

1. Agent usage is exploding
   Developers and operators are already running multiple agents at once: planning, coding, testing, incident response, and more. The problem is no longer whether agents exist, but how they work together at scale.
2. Workflows are crossing organizational boundaries
   Agents don’t just live inside one repo or one SaaS account anymore. They’re interacting across internal systems, external vendors, and partner environments. That requires a secure, programmable way to manage identity, permissions, and data flows between agents.
3. Enterprises are demanding governance, not demos
   Boards and CISOs are asking harder questions: Who approved this agent action? What policies apply? How do we track what happened? There is a need for a dedicated communication and control layer.

If the last decade was about building cloud infrastructure and the last few years were about models, the next decade will be about coordinating AI workforces.

BAND is positioned exactly at that inflection point.

Why this team?

Our conviction in BAND starts with the founders.

Our conviction in BAND starts with its founders. CEO Arick Goomanovsky previously co-founded our very own portfolio company, Sygnia (acquired by Temasek) and Ermetic (acquired by Tenable), building and scaling global enterprise security businesses, while CTO Vlad Luzin brings nearly two decades of distributed systems experience as VP of AI & Cloud Platform at Verint, senior product and engineering leader at CME Group, and former head of a Multi-Agent AI incubation team at Samsung Telecom Research. They’re not chasing a trend, they’re solving problems they’ve lived.

From our earliest conversations, a few things stood out:

• Clarity of the problem- Arick and Vlad could describe (in painful detail I must admit) how multi‑agent workflows break as they scale: lost context, brittle glue code, and humans stuck coordinating tools.
• Realism about enterprises- They aren’t trying to replace everything you already have. BAND is intentionally built to sit between existing agents and systems.
• Bias toward infrastructure, not features- Many agent products start as flashy interfaces and then add infrastructure later. BAND went the other way: start with a solid communication and governance layer, then let others build interfaces and agents on top.

As investors, we look for founders who are opinionated about the problem, humble about the unknowns, and relentless about execution and Arick and Vlad check all three boxes.

At Team8, our AI & software infrastructure thesis is simple: as AI moves from prototypes to production, the bottlenecks shift from “can the model do it?” to “can the system support it safely, reliably, and at scale?”

We’ve backed companies that form critical layers of this new stack, and BAND sits in that portfolio as the coordination layer for AI agents. In the near term, we expect to see the platform power agentic workflows in software engineering planning, coding, review, testing, and deployment, all coordinated across a mesh of specialized agents.

We’re proud to be partners to Arick, Vlad, and the entire BAND team as they build this new layer of the AI stack. You can learn more and start experimenting with the platform here: https://www.band.ai/

And yes,  we know investing in BAND means we’re officially in the business of backing great bands 🙂