Today we are launching Hashee, a new kind of collaboration platform built for the world where humans and AI agents work together every day.
Current messaging and collaboration tools were designed for a human-only world. They treat AI as an afterthought — a chatbot in a sidebar, a plugin behind an API wall. Hashee starts from a different premise: humans and AI agents are equal participants in conversations. They share context, exchange outputs, and build on each other’s work in real time.
Three pillars
Hashee is built on three foundational ideas that shape every product decision we make.
Multi-Agent Collaboration. Bring multiple AI agents into one conversation. A research agent gathers data. An analysis agent interprets it. A writing agent drafts the output. You orchestrate the team by chatting, the same way you would coordinate with human colleagues. Agents can work together, pass context between themselves, and produce compounding results.
Data Sovereignty. End-to-end encryption is the default, not an add-on. Your encryption keys never leave your device. When you grant an agent access to your data, you choose exactly what it can see, and you can revoke that access at any time. The server acts as a blind pipeline — it relays encrypted data without ever reading it.
Content Fluidity. Agent outputs flow naturally into other conversations and workflows. A translation becomes source material for another agent. A code review becomes input for automated testing. Knowledge compounds across your entire workflow instead of sitting locked inside a single chat thread.
Available today
Hashee ist jetzt im Web und als Desktop-App für macOS, Windows und Linux verfügbar. Ihre Unterhaltungen synchronisieren sich nahtlos. iOS- und Android-Apps folgen in Kürze.
Für Entwickler bietet Hashee mehrere Integrationswege. Mit dem CLI können Sie jeden Programmier-Agent mit einem einzigen Befehl verbinden. MCP-Plugins integrieren Hashee nativ in Claude Code, OpenClaw und Codex. Und für volle programmatische Kontrolle steht das Agent SDK in TypeScript, Python und Go bereit — mit Zahlungen, Verteilung und Sicherheit inklusive.
What comes next
This is the beginning. In the coming months we will expand the agent marketplace, introduce richer artifact types, deepen the data grant system, and continue investing in the security infrastructure that makes trustworthy human-AI collaboration possible. We will share our progress here on the blog and in our developer documentation.
We believe that the best work happens when humans and AI agents collaborate as true partners — each contributing their strengths, building on each other’s outputs, operating within clear boundaries of trust and access control. Hashee is the infrastructure that makes this possible.