Agent Landscape

Open-source personal agent software that made messaging apps a primary surface for autonomous workflows.

• Personal-agent runtime rather than a single chatbot: it combines long-running agent sessions, skills, memory, and user-facing communication surfaces.
• Messaging apps are documented as primary user-facing surfaces for operating the agent.

Nous Research agent with a gateway model for talking to one agent from chat and terminal surfaces.

• Gateway-style agent runtime that exposes one agent identity through chat apps, email, browser, API, webhook, CLI, and related control surfaces.
• The official Messaging Gateway docs list a broad set of platform adapters for reaching the same agent.

Minimal coding agent and agent stack used underneath OpenClaw-style channel-connected assistants.

• Small terminal coding harness associated with OpenClaw-style implementations rather than a consumer-facing messaging product.
• Official channel references point to a separate chat automation plugin rather than Pi core.

Ultra-lightweight open-source personal agent for tools, chats, memory, automation, model routing, and deployment.

• Official repository describes nanobot as a lightweight personal agent with WebUI, chat channels, tools, memory, MCP, model routing, automation, and deployment.
• Channel support follows the official chat-apps documentation and channel implementations.

Open-source personal assistant infrastructure for running autonomous agents across operating systems and deployment targets.

• Personal-assistant infrastructure aimed at portable, autonomous agents across OS and deployment environments.
• Channel details are limited to adapters documented by the official repository and docs.

Lightweight OpenClaw alternative that runs in containers and connects agents to messaging apps and scheduled work.

• Container-isolated OpenClaw alternative with memory, scheduled jobs, and explicit messaging-app connections.
• WhatsApp, Telegram, Slack, Discord, Gmail, memory, and scheduled work are all part of the public project description.

NVIDIA open-source runtime layer for running Hermes, OpenClaw, and related agents inside OpenShell with managed inference.

• Runtime/security layer for operating OpenClaw-family agents with managed inference rather than a chat surface by itself.
• Official materials describe runtime infrastructure, not a user-facing messaging agent.

Agent OS focused on privacy, security, and extensibility for OpenClaw-like personal-agent workloads.

• Agent OS rather than a wrapper: the stated axis is privacy, security, and extensibility.
• Official channel claims are narrower than the broader agent-OS positioning.

OpenClaw coworker-style agent project for turning conversations and delegated work into executed tasks.

• Coworker-style OpenClaw product framing: delegated work is the core metaphor rather than a generic UI layer.
• The official repository documents nanobot-based channel integration for the ClawMode workflow.

Self-evolving agent project where interaction and conversation are part of the agent's improvement loop.

• Self-evolving agent project centered on learning through interaction.
• Channel evidence is limited to optional WeChat integration and API/proxy client surfaces.

Go rebuild of OpenClaw focused on multi-tenant isolation, native concurrency, and safer runtime deployment.

• Runtime rebuild of OpenClaw in Go with concurrency, multi-tenancy, and security as the differentiators.
• Official materials describe both runtime infrastructure and user-facing messaging channels.

OpenClaw alternative with a web UI, sandboxed runtime, built-in IM support, and channel-based collaboration.

• Product-style OpenClaw alternative with runtime isolation, IM support, and channel-based collaboration.
• The public description includes built-in IM support, sandboxed runtime, and channel-based collaboration.

Source-available Go runtime for creating, managing, and running AI agents from a single binary.

• Agent factory runtime with per-agent personality, memory, skills, tools, session management, sandboxing, and OpenAI-compatible API support.
• It exposes deployable agent infrastructure and per-agent channel bindings rather than only a hosted end-user product.

BytePlus managed cloud service for running OpenClaw without operating the infrastructure yourself.

• Cloud-hosted OpenClaw service: the product value is managed deployment, hosted runtime, and operational packaging.
• Official materials describe it as a managed cloud service for OpenClaw-family agents rather than an independent desktop client.

Moonshot AI's browser-accessible Kimi surface for OpenClaw-style cloud agents and Claw Groups.

• Kimi-managed surface for OpenClaw-style cloud agents, with official documentation connecting Kimi model usage to OpenClaw workflows.
• Public official evidence for external messaging-channel routing is still limited.

Open-source WeChat bridge that turns WeChat into a control surface for OpenClaw and other local or remote agents.

• Channel bridge focused on making WeChat a control interface for OpenClaw-family and coding agents.
• It is a bridge layer for WeChat-based agent control rather than a full standalone agent product.

MarsWave agent operating system centered on memory, personality, judgment, and long-running user context.

• Agent OS product centered on persistent memory, personality, judgment, and continuity.
• Public materials emphasize agent continuity, native desktop operation, and Telegram/Lark/Cola messaging surfaces.

Baidu AI Cloud desktop and mobile agent for office automation, local files, software operation, and multi-step work.

• Desktop agent product for files, office workflows, software operation, and multi-step task execution.
• Official integration guides document Feishu, WeChat, DingTalk, and related Baidu work channels.