The Falconry
The Falconry is a GitHub organization/project built around Talon, an open-source, self-hosted harness for long-lived AI agents. Its stated philosophy is autonomy with a human remaining accountable: agents have persistent memory, tools, scheduled/background work, and a visible record of what they do.
At a Glance
AI Tools by The Falconry
(1)
Talon
Self Hosted Agentic AI Harness
Discussions
No discussions yet
Be the first to start a discussion about The Falconry
Latest News
Talon v5.39.5 released with current standalone binaries/packages and dependency updates including CodeQL action updates.
Talon v5.39.4 released as a patch release.
Talon v5.39.3 released with dependency/security updates, including follow-redirects and Hono updates.
Talon v5.39.0 released, followed by patch releases in the same release series.
Products & Services
Open-source, self-hosted multi-platform AI-agent harness distributed from GitHub, npm as talon-agent, and Homebrew. It connects interchangeable agent backends to chat, terminal, desktop/mobile, and device-mesh frontends.
Reference Flutter companion application for Windows, macOS, Linux, and Android, connecting to the Talon client bridge for multi-chat history, streaming, reasoning/tool visibility, configuration, and updates.
Built-in and external hot-reloadable plugins, including GitHub, MemPalace, Playwright, Brave Search, and organization repositories for Google, email, FFmpeg, Polymarket, FLI, Whois, X/Twitter, Codex, and extras.
Market Position
Talon positions The Falconry's work as a self-hosted, long-lived agent harness rather than a hosted chat product or one-shot CLI wrapper. Its differentiators are backend interchangeability, persistent memory, background autonomy, multi-channel frontends, hot-reloadable MCP tooling, companion apps, and operator control of data; likely comparison categories include hosted AI assistants, coding-agent CLIs, and single-channel bot frameworks.
Founding Story
The project frames its design around the old falconry partnership: the agent should fly and act with independent judgment, but return to a human who answers for its work. The initial vision is therefore a long-lived, checkable agent rather than a one-shot terminal wrapper or thin chat interface.
Business Model
Revenue Model
The available sources describe Talon as Apache-2.0 open-source software distributed through GitHub, npm, and Homebrew; they do not describe a paid subscription, API-usage charge, or commercial licensing program. Users bring their own model/backend credentials or local services.
Pricing Tiers
Apache License 2.0; installation is documented through npm, Homebrew, source checkout, and downloadable binaries. Users may incur separate costs from third-party model providers or services they choose.
Target Markets
- Running a continuously available personal AI agent on a self-controlled VPS or computer
- Conversational agent access through Telegram, WhatsApp, Discord, or Microsoft Teams
- Terminal and developer workflows with MCP tools, GitHub access, coding backends, and sub-agents
- Long-running goals, scheduled jobs, proactive heartbeat maintenance, and memory consolidation
- Remote administration and action across paired personal devices through the device mesh
- Building or extending an agent with custom MCP plugins, skills, triggers, or frontends