NEWS · JULY 21, 2026 · AGENTIC AI

Agents move into the team's chat channel

Block, Jack Dorsey's company, introduced Buzz, a group chat app built so teams and AI agents can work in the same workspace. Dorsey describes it as model-agnostic, decentralized and open source. A free desktop app is available for macOS, Windows and Linux. This issue covers the July 21 agenda and was published on July 25, 2026 to complete the archive.

01 · WHAT HAPPENED?

It looks like Slack, but with agents inside

Block, the company behind Square, Cash App, Afterpay and Tidal, released Buzz. The app aims to combine team conversation with AI agents in one workspace: people and agents can work together inside the same thread, reducing the need to move between tools. The interface resembles Slack, but it differs by building agent integration in natively and pulling GitHub project management into the same window.

02 · DETAILS

Open source, model-agnostic, early stage

Dorsey describes the app as model-agnostic, decentralized, self-sovereign and open source. The source code is available on GitHub, and developers with full access can spin up their own customized instances. The desktop app is offered free on macOS, Windows and Linux.

That said, the product is at an early stage, which calls for caution before migrating. A team communication tool should not become an organization's main channel before its reliability is proven. The approach itself is still notable, because it places agents not in a separate interface but where the work is already being discussed.

03 · WHY IT MATTERS

An agent belongs where the work is discussed

Most AI tools live in a separate tab: the work is discussed in one place, asked about in another, and the result carried back by hand. Placing agents directly in team chat claims to remove that break. When the context stays in one channel, the team can also see what the agent based its decision on. Being open source and model-agnostic sends a second message: organizations want to decide both which model they use and where their data sits. That suggests locking into a single provider is starting to be seen as a risk.

04 · TÜRKİYE

What it means for businesses in Türkiye

Most small and mid-sized teams in Türkiye run work conversations through WhatsApp groups. It is a practical solution with two problems: personal and business communication blur together, and knowledge leaves with the employee who held it. What tools like Buzz point to is team communication that is both on record and connectable to automation. A business does not need to move to Buzz today; making an early-stage product your main channel would be risky. But the question is clear: are your work conversations somewhere searchable and transferable, or in a phone contact list? For teams considering agents, a second question follows: what information circulates in the channel the agent can see, and is processing it appropriate under KVKK? The boundaries of that conversation need drawing before the agent is invited into it.

The UNALSOFT take

Our reading is this: the easiest way to bring agents into the work is not buying a new tool but placing them inside the existing flow with the right boundaries. In a team where nobody knows what data moves through which channel, an agent produces more risk than productivity. That is why we keep the order intact in our Agentic AI service: process and data boundaries first, agent second. Tool choice comes last, and it is the piece most easily changed anyway.

Draw the boundary before inviting the agent.

Let's map what data moves through your team communication and what could be automated.

Message on WhatsApp