Dots, Instinct and Muse: Always-On Agents Aimed at Different Customers

A shared category, split by audience

Schario describes always-on agents as products that understand what you are working on, retain context and take action on your behalf without needing you to direct every step. She notes that OpenClaw helped popularize the category by showing what an agent could look like when it ran continuously and could be reached through messaging apps people already use. According to the post, that project started in November 2025.

The biggest difference she sees between the three newer products is the customer. Dots is positioned for professional use. OpenAI demoed it fixing software, updating launch materials, working through research and preparing sales proposals. Instinct is built around personal use cases such as booking a ride to the airport, finding a handyman and following up on conversations you have dropped. Meta's Muse is also positioned as a personal agent.

Schario says this reflects each company's go-to-market. OpenAI is building an enterprise business, while Meta is a consumer technology company. She stresses that this is her interpretation of the positioning and that the capabilities overlap. In her view, the examples a company picks show which customer it wants to reach and what it wants that customer to try first.

Where you talk to the agent

The products also differ in how you interact with them. Dots is used through ChatGPT, with Slack and Teams available as additional channels. Instinct works through iMessage or WhatsApp. Muse has its own app and is also available through WhatsApp.

Schario argues that an agent's location shapes what people ask of it. An agent in Slack will likely be pulled into work conversations. One in your messages can be texted when you remember an appointment to make or something to book for tomorrow. For personal use, she says, being able to text an agent means you do not have to remember to open another app. For professional use, being reachable in tools a team already uses serves a similar purpose.

Credentials, training and advertising

An always-on agent needs access to be useful, which raises separate questions: can the model see your passwords, can the company train on your information, and can it be used for advertising. Schario says the answers vary by product.

• Dots: OpenAI says supported secure login forms keep credentials from being exposed to the model, but this does not apply to passwords typed directly into a chat, document or plugin. Business, Enterprise and Edu data is not used for training by default. On personal plans, the model-improvement setting decides whether eligible conversations and work can be used. OpenAI also says it does not train directly on proactive background research or the agent's notes, though information from that research can become eligible if used in a conversation or task. Dots is currently offered on ad-free plans, which Schario says is not the same as an explicit commitment to keep agent data separate from advertising systems.
• Instinct: Its privacy policy anticipates users providing usernames and passwords so it can sign into accounts. Schario did not find a comparable public technical explanation of how those credentials are kept separate from the model. Instinct allows information to be used for training, with an opt-out and some safety-review exceptions. It says it does not use Google Workspace API data for model training or advertising, which Schario describes as a specific commitment rather than a blanket one.
• Muse: Meta says passwords and authentication tokens are stored separately from the agent, which can use them without seeing the real credentials. Meta also says Muse conversations and data in its virtual machine are not shared with its advertising systems. However, if Muse visits a retailer's website or acts on another service, that activity can still affect the ads you see. Meta says sanitized interaction data is used for training unless you opt out, and its launch architecture allows Meta to access data when necessary to operate, support or secure the service.

Schario points out that these are the companies' published descriptions and policies. They show what each company commits to, but do not independently establish how well the protections work.

What to compare before choosing

Schario expects the professional and personal use cases to overlap more over time, since a business trip involves both work and personal planning, and an inbox holds both kinds of tasks. For now, she suggests comparing the agents on three points: what they are built to help with, how you interact with them, and what happens to the access and information you give them. Her advice up front is to try them all.