Buzz vs Kylon vs Slack AI: Which AI Workspace Is Right for Your Team?

Three platforms. One premise: AI agents should work alongside humans, not just answer their questions.

Buzz, launched by Jack Dorsey's Block on July 21, 2026, is the newest entrant. Kylon has been building the AI-native workspace for over a year. Slack added AI features and Agentforce integration to the platform 42 million people already use.

Each one approaches the human-agent collaboration problem differently. This comparison helps you decide which architecture fits your team.

Quick comparison

Agent architecture

This is where the three platforms diverge most sharply.

Buzz: cryptographic agents for code

Every Buzz participant, human or agent, gets a secp256k1 cryptographic keypair via the Nostr protocol. This means agent identity is portable, verifiable, and not locked to any platform. Agents sign their own work, and a secondary signature ties each agent back to the human who authorized it.

Block's engineering blog describes an impressive pattern: one frontier agent driving a swarm of cheaper, faster agents. The smart one plans; the fast ones research, build, test, and review in parallel. They even "invent coordination we didn't script: recruiting each other, splitting work into side channels, handing tasks across contexts."

The limitation: every example in Block's documentation is a software development scenario. Code reviews, CI pipelines, branch management, release automation. The agent architecture is deeply tied to git workflows.

Kylon: agents as team members

Kylon agents are persistent members of your workspace. Each one has:

• Identity that persists across every conversation
• Memory that grows and compounds over weeks and months
• Skills that encode domain expertise (brand guidelines, SEO auditing, TikTok analytics, financial analysis)
• Permissions scoped to what they should access
• Connections to external tools through OAuth (Gmail, GitHub, Notion, Google Search Console, Ahrefs, Meta Ads, PostHog, and more)

Multiple agents collaborate in shared channels. A marketing agent runs ad campaigns, a research agent monitors competitors, a content agent writes blog posts, and a deployment agent publishes them. Each one maintains its own context, but they all share the same workspace and can hand off work to each other.

The key difference from Buzz: this works for marketing, sales, finance, operations, and support, not only engineering.

Slack AI: AI features on existing infrastructure

Slack took the opposite approach. Instead of building a new workspace, it added AI capabilities to the platform millions of teams already use:

• Conversation summaries that condense long threads
• Daily recaps of what you missed
• Slackbot as a personal AI agent for Q&A
• AI workflow generation from natural language
• Enterprise search across connected apps (Business+ and above)
• Agentforce agents that can be @mentioned in channels (requires Salesforce)

The advantage is adoption: no migration needed. The limitation is depth. Slack AI does not maintain persistent agent identity or memory across conversations. Without Salesforce, you get summarization and search but not autonomous agents.

Business breadth

Buzz is explicitly built for software teams. Block's README lists three example stories: incident response, feature branch management, and release automation. There is nothing wrong with this, but it means Buzz will not help your marketing team run campaigns or your finance team build dashboards.

Kylon is designed for every department. The same workspace handles SEO workflows, ad campaign management, sales pipeline tracking, competitive intelligence, content production, and internal data applications.

Slack AI serves every department superficially (summaries and search work everywhere) but deep autonomous agent capabilities require Salesforce Agentforce, which adds cost and complexity.

Integrations

Buzz is git-native. Code repositories, branches, commits, PRs, and CI are first-class concepts. For everything else, agents need to use the Agent Client Protocol (ACP) to connect to external tools. The ecosystem is young.

Kylon has 3,000+ integrations via OAuth and MCP that agents can use with proper permission scoping: Gmail, Outlook, GitHub, Notion, Slack, Google Search Console, Ahrefs, Meta Ads, Google Ads, PostHog, HubSpot, Apollo, and more. Agents also have built-in capabilities for web search, web scraping (via Apify and Bright Data), image generation, video creation, voice calls, and browser automation.

Slack has the largest app directory (2,600+ apps), but most are human-operated integrations, not agent-accessible tools. Agentforce agents can take actions across Salesforce products, but connecting to non-Salesforce tools requires custom development.

Self-hosting and open source

Buzz is fully open source (Apache 2.0). You can run your own relay on your own infrastructure. This matters for teams with strict data sovereignty requirements or those who want to audit every line of code their workspace runs.

Kylon is cloud-only. The platform handles hosting, scaling, and security. SOC 2 compliance is in progress.

Slack is cloud SaaS owned by Salesforce. Enterprise Grid offers data residency options, but self-hosting is not available.

If self-hosting is a hard requirement, Buzz is your only option on this list.

Pricing

Buzz's app is free and open source. But you bring your own model API keys (Claude, GPT, Codex), so the real cost is API usage plus the infrastructure and engineering time to set up and maintain a relay. Kylon is also free to start with a 7-day trial and usage-based pricing after that. For teams without DevOps capacity, the operational overhead of self-hosting Buzz can exceed the cost of a managed platform like Kylon.

Who should choose what

Choose Buzz if:

• Your team is primarily software engineering
• You care deeply about open source and self-sovereignty
• You have the DevOps capacity to run your own relay
• Decentralized identity (Nostr) matters to your architecture
• You want to audit every line of code in your workspace

Choose Kylon if:

• You need AI agents across multiple departments (marketing, sales, finance, operations, engineering)
• You want agents that maintain persistent memory and context across months of work
• You need native integrations to business tools (CRMs, ad platforms, analytics, email)
• Your team needs to build internal data apps alongside agent workflows
• You want a managed platform without self-hosting overhead

Choose Slack AI if:

• Your team already lives in Slack and switching is not an option
• You need AI summaries and search but not autonomous agents
• You are already on Salesforce and want Agentforce integration
• Incremental improvement over your current workflow is more important than architectural change

The bigger picture

Buzz, Kylon, and Slack AI represent three different bets on how work will change:

• Buzz bets that the future of work is open-source, decentralized, and engineering-first. Build the protocol, and the ecosystem will follow.
• Kylon bets that every team needs AI teammates, not just engineering. The workspace itself should be rebuilt around human-agent collaboration.
• Slack bets that AI should be added to where people already work. Don't change the workspace; change what the workspace can do.

All three bets have merit. The right choice depends on your team's size, technical capacity, departmental breadth, and willingness to adopt a new platform versus improving an existing one.

This comparison is based on publicly available documentation, engineering blog posts, published pricing, and hands-on testing as of July 2026. We build Kylon and have presented each platform's genuine strengths alongside its limitations. Corrections welcome.

Last updated: July 28, 2026