A self-hosted agent across every chat, or a system that acts on what your team receives.
OpenClaw is the dominant personal AI agent on the market by traction. Open-source, MIT-licensed, runs on your own machine, lives in every chat app you already use (WhatsApp, Telegram, Slack, Discord, iMessage, Teams, and dozens more), with persistent memory kept locally. You message OpenClaw from any chat app to ask it to do something, and it does it. this+that is different. It's a comms-forward AI that reads the messages your team receives, scans every email and chat for what needs doing, surfaces tasks for the whole team, and runs workflows. Two different shapes of AI for work, not really direct alternatives.
A self-hosted personal agent, or a comms-forward AI for your team's messages
AI assistants come in different shapes. We think of three waves: Wave 1 is the general-purpose chat you open and ask (Claude, ChatGPT, Gemini, Copilot). Wave 1.5 is the polished personal agent layered on top of chat. Wave 2 is the fit-for-purpose AI that owns one job end-to-end, with UI tailored to it (more in our essay on AI's second wave).
OpenClaw sits in Wave 1.5, the polished personal-agent tier between raw chat and fit-for-purpose Wave 2 products. It is the dominant Wave 1.5 player by a wide margin: hundreds of thousands of GitHub stars, hundreds of contributors, broad community adoption, and stewardship by an independent foundation that OpenAI publicly supports after creator Peter Steinberger joined OpenAI in February 2026. The product wraps general-purpose models in a routing layer that lets you talk to one agent from every chat app you already use, plus a local-first memory layer that stays on your machine. It is closer to fit-for-purpose than vanilla chat, but it is still chat-shaped at the core. You still go to it and ask. this+that is one of Wave 2's first products. The agent reads the messages your team receives whether you asked for them or not. Every Gmail or Outlook message, Slack thread, Teams chat, Google Chat, and Telegram DM gets scanned for what needs doing, surfaced as tasks for the whole team, and runs through workflows. Most people who use OpenClaw would also benefit from a Wave 2 product handling their team's inbox. They solve different problems.
Feature by feature
How the two products compare across the things that matter most.
| Topic | OpenClaw | this+that |
|---|---|---|
| Category | Wave 1.5. Self-hosted, open-source personal agent that lives in your chat apps. Same shape as Town, Manus, and Hyperagent, with more traction than the others. | Wave 2. Fit-for-purpose AI that lives inside the work surface, acting on messages that arrive without you asking. |
| Where work starts | With you. You DM OpenClaw from any chat app to ask it to do something. Cron jobs, reminders, and proactive heartbeats add a layer of scheduled actions. | With incoming messages. The agent reads what arrives without being asked and surfaces what needs doing. |
| Hosting model | Self-host (MIT licensed). Managed OpenClaw Cloud reported at $59/mo (first month $29.50). Third-party managed hosts available. | Hosted SaaS. No setup, no servers to run. |
| Channels | 20+ chat surfaces: WhatsApp, Telegram, Slack, Discord, Google Chat, Signal, iMessage, Teams, Matrix, Feishu, LINE, Mattermost, IRC, and more. Plus CLI, macOS/Windows companions, voice, and a Canvas surface. | Gmail, Outlook, Slack, Teams, Google Chat, and Telegram. Every message scanned for what needs doing. |
| Task extraction from messages | Not the focus. The agent acts when you DM it or when a scheduled heartbeat fires. Messages arriving for you are not scanned for tasks. | Core capability. DoBox scans every message across every channel for requests, commitments, deadlines, and follow-ups, surfaced as assignable tasks for the team. |
| Memory / knowledge | Persistent personal memory stored locally on your machine. "Remembers you and becomes uniquely yours." Single-user. | The brain. A writable team knowledge layer with personal and team spaces. Agents read and write. Pages can be published outward through public artifacts. |
| Workflows / routines | Cron jobs, reminders, background tasks, and proactive heartbeats. Per-channel multi-agent routing isolates sessions. | Plain-language workflow builder. 32 action types. IF/ELSE branching, loops, and sub-agents. Triggered by messages or schedules. |
| Team capabilities | Single-user focused. Memory and config live on the operator's machine. | Team task assignment, shared task view, and team-request grounding (your agent searches a teammate's brain to draft a suggested reply). |
| Trust controls | DM pairing: unknown senders are blocked until approved with a pairing code. Opt-in auto mode for enterprises routes higher-risk actions to human review. | Approval gates on every workflow action. Drafts not sends by default on outbound. Provenance: every task links back to the source message. |
| Integrations | MCP Registry for external tools plus a community skills system. Full system access on the host machine (browser, files, shell). | Open MCP standard. Any tool with an MCP server or API plugs in. |
| Pricing | Free to self-host. Managed OpenClaw Cloud reported at $59/mo (first month $29.50); third-party managed hosts ~$45-55/mo. Verify pricing on openclaw.ai before publishing. | Free during beta. |
- OpenClaw is the dominant Wave 1.5 personal agent by a wide margin. Hundreds of thousands of GitHub stars, hundreds of contributors, a strong community, and momentum the rest of the tier does not match.
- Self-hosted and MIT-licensed. If keeping data on your own machine is a hard requirement, OpenClaw is one of the few serious options.
- The chat-surface coverage is genuinely impressive: WhatsApp, Telegram, Slack, Discord, iMessage, Teams, Matrix, and a long tail of others. One agent, every place you already talk.
- Local persistent memory is a real personalization layer for individual use, with no vendor cloud involved.
- DM pairing is an elegant trust model for an agent that lives in unrestricted DM surfaces, and the opt-in enterprise auto mode is sensible.
- Open foundation governance and OpenAI support post-Steinberger give the project unusual long-term stability for an open-source personal-agent project.
- A different category. The agent reads the messages you and your team receive across email, Slack, Teams, Google Chat, and Telegram, and acts on them without you asking.
- DoBox: every message scanned for tasks, with the whole team able to see, assign, and track them in one place.
- Plain-language workflow builder triggered by incoming messages. 32 action types, branching, loops, sub-agents.
- The brain is a writable team knowledge layer (not single-user local memory) that agents read and update as a side effect of doing real work.
- Hosted SaaS, free during beta. No machine to run, no cloud to manage.
- Team-first design: task assignment, shared views, and team-request grounding are first-class capabilities, not personal features.
See what's waiting in your inbox
Connect your inbox and get a free analysis of the tasks, follow-ups, and open commitments this+that would surface across your channels.
Free to use · No credit card required