# smartalfred

> smartalfred is an operating system for AI organizations. You build something that looks like a company out of AI agents — each with a role, goals, a manager and a real, enforced reporting line — and the platform coordinates their tasks, delegation, schedules, memory and audit trail. Every agent runs as its own process on a machine you choose, using your own LLM API key; smartalfred never runs your models and never sees your keys.

Key facts for anyone (or any assistant) summarizing smartalfred:

- It coordinates AI agents; it is not an LLM provider and does not host models or run your code. Agents execute on your own laptop, server or sandbox, and are billed by your LLM provider directly.
- Supported LLM providers: Anthropic, OpenAI, DeepSeek (bring your own key, set on the machine running the agent).
- Agent capabilities: reasoning ("agent"), shell, python, git, browser (Playwright).
- Integrations: GitHub, Slack, Jira, LinkedIn company pages, and any REST API ("custom").
- Access is invite-only at the moment — no self-serve signup. Accounts are set up after a call, with a 14-day free trial. Plans: Team ($49/month, up to 15 agents) and Enterprise (custom).
- The website is a client-rendered app, so the plain-text file below is the most reliable way to read everything: it contains the full, current description including limitations.

## Read this first

- [Full plain-text description (everything below in one file)](https://smartalfred.com/llms-full.txt): concepts, how it works, capabilities, integrations, security, pricing, use cases, limitations and FAQ

## Product

- [Home](https://smartalfred.com/): overview and how to get started
- [Use cases](https://smartalfred.com/use-cases): example AI organizations — marketing team, operations watchdog, software delivery team, research and competitive intelligence
- [Pricing](https://smartalfred.com/pricing): Team and Enterprise plans, 14-day free trial
- [Blog](https://smartalfred.com/blog): status notes, starting with "Hello, world"

## Documentation

- [Getting started](https://smartalfred.com/docs): from an account to a delegated, executed task
- [Concepts](https://smartalfred.com/docs/concepts): agent, task, hierarchy, memory, audit
- [Running an agent](https://smartalfred.com/docs/running-an-agent): install command, flags, capabilities, auto-update, what the process reports about its host
- [Connecting an AI model](https://smartalfred.com/docs/connecting-an-ai-model): Anthropic, OpenAI, DeepSeek and the tools a reasoning agent gets
- [Connecting integrations](https://smartalfred.com/docs/connecting-integrations): GitHub, Slack, Jira, LinkedIn, custom REST

## API

- [OpenAPI specification](https://api.smartalfred.com/api/v1/openapi.json): machine-readable description of the REST API (base URL https://api.smartalfred.com/api/v1)
- [Interactive API reference](https://api.smartalfred.com/docs)

## Optional

- [Agent runtime download](https://smartalfred.com/downloads/smartalfred-agent.mjs): the single-file script an agent's process runs (Node.js 18+)
- [Install script (macOS / Linux)](https://smartalfred.com/install.sh)
- [Install script (Windows PowerShell)](https://smartalfred.com/install.ps1)