What Olano is

Olano runs agents: long-lived AI workers that hold their own memory, reach real systems through tools, and talk to people on the messaging platforms those people already use.

An agent is four things:

Piece	What it is
Configuration	agent.json5 — the agent's name, model, prompt, tools, channels, schedule, permissions.
Workspace	A directory the agent owns: its memory, notes, files, skills, and scratch space.
A model	The reasoning engine. Either an Olano-managed tier billed in credits, or your own provider key.
Surfaces	How work reaches it: the dashboard, a messaging channel, a schedule, a webhook, or another agent.

The pieces

• The runtime is one long-running process. It boots every agent, keeps the channels connected, runs the scheduler, and serves the dashboard API.
• The dashboard is the web control panel — the primary way to operate a deployment. Everything in this guide that mentions a "tab" means a dashboard section.
• The agents themselves. Every agent carries built-in tools to configure itself, connect services, and send magic links so a person can finish a sign-in or hand over a key from their own phone; AgentFather carries the tools to manage the whole deployment. Asking an agent is a first-class way to operate Olano, not a shortcut — see Ask the agent: self-configuration, magic links and admin tools.
• The vault is the encrypted credential store. Every API key, bot token and service password lives there. Credentials never belong in a config file.

Where things live on disk

Olano keeps everything under one home directory, ~/.lionclaw/ by default (override with OLANO_HOME):

~/.lionclaw/
  config.yaml                       # deployment-wide settings  (config.yaml reference)
  organization.yaml                 # the org chart             (organization.yaml)
  agents/<name>/
    agent.json5                     # that agent's config       (agent.json5 reference)
    workspace/                      # that agent's own files
      AGENTS.md                     # operating instructions
      SOUL.md                       # persona, tone, boundaries
      IDENTITY.md                   # who the agent is
      USER.md                       # who it works for
      TOOLS.md                      # notes on tool usage
      MEMORY.md                     # durable learned facts
      memories/YYYY-MM-DD.md        # daily logs
      notes/                        # the markdown wiki
      knowledgebase/                # documents indexed for search
      skills/                       # reusable procedures
      reports/                      # every file the agent produces for you
      uploads/                      # files you send it in chat
      downloads/  scratch/  trash/  # conventions the tools use
      system/                       # Olano-owned; replaced on upgrade
  sessions/                         # conversation transcripts
  vault/                            # encrypted credentials

The files under system/ are managed by Olano and overwritten when you upgrade. Everything else is yours — edit it freely, from the dashboard Files tab or a text editor. The agent reads AGENTS.md, SOUL.md, IDENTITY.md, USER.md, MEMORY.md, TOOLS.md and the recent daily logs at the start of every conversation, so editing them changes how the agent behaves on its next reply.

Self-hosted or Olano Cloud

Olano runs the same either way:

• Self-hosted — you install it on your own machine or server and bring your own model provider keys.
• Olano Cloud (cloud.olano.ai) — a managed instance you deploy in a few minutes, kept up to date for you, with Olano-managed models billed in credits so you never need a provider account. See Olano Cloud.

You can mix: a cloud deployment can still use your own provider keys, and a self-hosted install can use whatever providers you have.