Muse, Instinct, Wajo, and OpenAI's Dots: Four AI Assistants That Do the Work for You

This week Meta released Muse, a cross‑platform assistant that runs on the company’s Muse Spark model and persists in a private virtual machine, while Stripe backs its checkout with a one‑time‑use card and warranty. At the same time, Instinct, a startup founded by 23‑year‑old Noah Shinn, closed a $1 billion Series C round that values it at $10 billion after a $250 million Series B, and it now offers a text‑or‑call interface that lives on a dedicated cloud computer for multi‑day jobs. Wajo introduced Fo, an “action agent” that can place outbound calls, hook into Google Workspace, and hire human assistants when needed, all for free. OpenAI’s Dots, powered by the upcoming GPT‑6 Astra, arrived as part of ChatGPT Pro and Business Premium, giving each dot its own cloud instance, read‑only access to 4,000+ plugins, and granular approval rules that separate credential handling from the language model. Together they represent the first wave of assistants that not only generate a plan but also execute it, asking only for user sign‑off on high‑impact decisions.

The launch of these four agents underscores a rapid escalation in the autonomous‑agent market. Meta and OpenAI bring deep pockets and platform reach, while Instinct and Wajo illustrate how lean startups can attract billion‑dollar valuations by solving the “execution gap.” All four rely on deep integration with personal data—email, calendars, payment accounts—and each adopts a different trust model: Muse isolates each user in a VM and uses Sentinel to vet outbound traffic; Instinct added a deletion tool after privacy concerns; Dots stores passwords in encrypted vaults and enforces rule‑based approvals; Fo leans on explicit user consent before major actions. The competition forces each player to differentiate through features such as outbound calling (currently unique to Fo), payment guarantees (Muse’s Stripe warranty), or the breadth of plugin connectivity (Dots).

Looking ahead, the biggest risks revolve around data privacy, inadvertent financial actions, and regulatory scrutiny of agents that can act on a user’s behalf. Users will likely start with low‑stakes tasks while testing each system’s permission model, and developers will watch for how quickly Muse adds outbound calling and whether OpenAI expands Dots beyond the ChatGPT ecosystem. The next inflection point may be enterprise adoption, where “specialist dots” with dedicated credentials could automate internal workflows, but success will hinge on transparent audit trails and robust revocation mechanisms.