Value Proposition & Features
Paseo is a self-hosted daemon for AI coding agents that lets users run agents on their own machine while connecting from phone, desktop, web, or CLI.
[1]
[2]
Its core value proposition is unified control of multiple agent providers—Claude Code, Codex, GitHub Copilot, OpenCode, and Pi—without moving code or credentials off the user’s local environment.
[1]
[3]
Core features include a client-server architecture centered on a local daemon and WebSocket connections for live agent management and streaming.
[1]
[4]
It also supports voice interaction , multi-device access , and privacy-first operation with no telemetry or forced logins, while keeping execution in the local shell environment.
[1]
[5]
[6]
Product Roadmap / Announcements
As of 2026-08-02, public update items surfaced in the official changelog include “Fast mode for Claude Opus” and “Multilingual local dictation with the new Parakeet v3 speech model.”
[8]
- 2026-07-05: Gigazine described Paseo as a free, self-hostable, open-source app for managing coding AI agents from smartphone, desktop PC, or terminal. [6]
Recent Developments
In the past 90 days, third-party coverage highlighted Paseo’s support for remote control of AI coding agents from multiple devices and emphasized encrypted relay access plus local voice processing.
[6]
The same coverage also noted CLI support for starting, listing, connecting to, and adding agents.
[6]
History and Origin Story
Paseo appears to have emerged as an open-source, self-hosted control layer for coding agents, with its official repository and documentation centered on a local daemon and multi-client architecture.
[1]
[7]
Available sources do not provide a clear founding date, founder names, or a detailed company origin story, so no reliable source found for those specifics.
[1]
[8]
Market Sizing
Category, Market Size, and Category Growth
Paseo fits best in the AI coding agent orchestration and developer tooling categories, specifically as a self-hosted control plane for agentic coding workflows.
[1]
[3]
The returned sources do not include credible analyst estimates for market size or growth, so no reliable source found for quantified sizing.
[1]
[8]