toonetooneThe knowledge is all there — in files, tools, and heads. Toone gives it structure: a living graph of what your company knows, and an organization of AI agents that puts it to work on schedule.
See how it worksSeven things become true once your organization runs on Toone.
Playbooks that run themselves as routines, sub-routines and schedules. Crafting one is deterministic work, much like wiring an n8n flow, except you write it in natural language. Every run stays debuggable, with agents diagnosing how it is going while it happens.
An organizational knowledge graph that outlives every chat and every employee.
They hold roles in departments, keep schedules, produce artefacts, and escalate to a human when it matters.
Routines are written to delegate. Work moves between departments in clean compartments, so every agent holds a narrow remit and the whole organization behaves like one body.
Your external tools are mapped as replayable workflows and treated as first-class integrations, scoped to what your company actually needs. Anything missing gets built custom.
Gates, approvals, drafts-only outbound and redaction. Autonomy inside the boundaries you set, with accountability for every action an agent takes.
Recommended templates, MCP tools, processes and integrations, assembled for your use case. Today that covers software engineering, SEO and growth, outreach and support, product strategy, science and research, and headquarters.
The short version of everything people ask us.
A macOS app that runs an organization of AI agents for your company — departments, agents with roles, and scheduled routines, built on a knowledge graph of what your company knows.
A chatbot answers when you ask. Toone's agents hold roles, follow routines on a schedule, share memory through the knowledge graph and hand work to each other — finished work arrives without prompting.
Recurring jobs — digests, content pipelines, monitoring, research — defined in plain files that agents execute on schedule, producing artefacts you can read and audit.
No. Organizations start from templates and routines are plain-language files. If you are technical, everything is inspectable down to the file level.
On your Mac. Your organization is a folder of readable files and agents run locally — only the model calls leave your machine.
Anthropic's Claude family, running through Claude Code. Each conversation thread can pin its own model.
The desktop app is macOS today, distributed as a direct download. More platforms are on the roadmap.
Toone is free to download during early access. Agents run on your own Anthropic subscription.
Yes — Toone white-labels: the same engine with your brand configuration.
