What is aouo?

aouo is an attempt to define the missing app layer for AI.

Most AI products today start from a chat box. That works for general assistance, but it is a weak shape for long-running, domain-specific work. A useful agent needs more than a system prompt. It needs state, schedules, tools, permissions, structured storage, and a way to be installed, inspected, upgraded, shared, and deleted.

The unit aouo defines is the pack: a local-first agent app bundle.

pack = skills + storage + memory + schedule + permissions + UI surface

A pack can be an English tutor, a writing assistant, a journaling companion, a research workflow, a content pipeline, or any other narrow agent that should remember things over time and run proactively.

The Host Model

aouo has three parts. They are intentionally separate.

aouo runtime
  Loads packs, runs agents, manages SQLite, memory, tools, schedules,
  permissions, usage, logs, and optional adapters.

aouo desktop / dashboard
  Lets users install, configure, chat with, inspect, debug, and control packs.

.aouo pack
  The installable AI app bundle: skills, memory defaults, optional schema,
  persist contract, UI surfaces, workflows, permissions, and evals.

The closest analogy is not “one chatbot with many prompts.” It is closer to:

VS Code  -> extensions
Obsidian -> plugins
Raycast  -> extensions
Docker   -> containers

aouo    -> .aouo packs

The runtime is the host. The pack is the app. The dashboard is where the two meet the user.

→ Read more in The Host Model.

What this is for

A pack is judged by whether a real user keeps coming back to it for weeks. Three concrete pack shapes the format is designed to support:

Pack	One-line scenario
English Coach	”Remember what I got wrong last week and quiz me on it this morning.”
Creator Studio	”Capture this link, draft three angles for it, slot the best one on Thursday.”
Daily Journal	”Two weeks of notes — what kept coming up?”

Each one has its own database, memory, cron schedule, skills, and permission scope. The runtime keeps them out of each other’s heads.

→ See Example Packs for these in depth.

Why “pack” and not “skill” or “GPT”

Generalist assistants, Custom GPTs, and flat skill libraries all hit the same wall: as the agent gets more capable, its context gets more polluted, its memory gets vaguer, and the user loses control of what it knows.

A pack draws a hard boundary at the runtime level:

• Conversation routing scopes every turn to one active pack
• Skills are namespaced by qualified name (<pack>:<skill>)
• Session keys mint a fresh history when the user switches pack
• Tool calls carry the active pack so the wrong pack literally cannot write to the right pack’s tables

This is what makes “installable AI apps” different from “one assistant with knowledge files.”

→ Read more in Why Packs.

What ships today

The current repo is a pre-alpha implementation of the runtime idea. It already contains:

• a pack loader and manifest validator
• pack-scoped SQLite databases
• pack-scoped memory files (soul.md / user.md / state.md)
• a ReAct-style agent runtime with quota gates and context compression
• provider integrations (Gemini, OpenAI, DeepSeek, Codex OAuth)
• a scheduler for proactive pack runs (cron)
• a Telegram channel adapter (voice, photo, document, streaming edits, forum topics)
• a local dashboard for configuration and inspection
• sample packs in apps/ for notes, content creation, and vocabulary learning

These pieces prove the runtime shape. They are not the final product surface — the next step is to make local usage feel first-class.

Where the product is going

near-term
  Expand the local dashboard into a real pack workspace
  (memory editor, DB browser, schedule panel, permission inspector)
  Local pack chat — no Telegram bot required
  Pack import / export

mid-term
  Package the local workspace as a desktop client
  Sharpen one flagship pack until it proves recurring use
  Pack builder — assemble packs from a natural-language brief
  Per-pack context budgeting visible and tunable

long-term
  .aouo as an open format for stateful agent apps
  Sharing, forking, auditing, upgrading packs safely
  More channel adapters as optional remotes (Discord, Slack)

→ Read more in The Desktop Direction and The Builder Direction.

Read next

If you’re orienting:

• The Host Model — runtime / desktop / pack split
• Why Packs — what the pack boundary buys
• Example Packs — three flagship pack shapes
• Context Compiler — how pack scope translates to smaller prompts
• Security & Trust — permissions, audit, sharing

If you want to build a pack:

• Your First Pack — tutorial
• Pack Spec — manifest fields
• Schema & Persist — storage levels and persist contract
• Skills — SKILL.md format

If you want to run the runtime today, it ships on npm as @aouo/agent — see the package README for install + CLI details.