· via TechCrunch
Superhuman pushes its Go AI assistant into Mail, Docs and 20 new agent integrations
Superhuman, formerly Grammarly, is embedding its Go assistant across Mail and Docs, opening it to 20 third-party agents through an SDK, and says Go has passed 1 million active users.

Go becomes Superhuman's default assistant
Superhuman, the company formerly known as Grammarly, is extending its Go AI assistant into every corner of its product line, according to TechCrunch. The rollout includes a more personalized Go assistant, native agents inside Docs, and integrations with third-party agents built on a new SDK.
The push began after the company rebranded last year, TechCrunch reports, and the latest update is designed to help Go learn context across different apps and offer better personalized suggestions for prompts. As part of that effort, Superhuman is reworking its onboarding flow so users can connect their preferred tools and load information about their work upfront.
Inline, proactive and style-aware
Like the writing checks Grammarly was known for, Go operates inline with the cursor, but it also manages other agents. TechCrunch describes the assistant as proactive: it can check your schedule, monitor your email and draft responses on its own. It also learns individual writing styles and applies them when composing in email and Docs.
Docs and Mail rebuilt around agents
In Docs, all Go agents now run natively. Users can set them loose on different tasks in the background and configure routines that trigger workflows based on events.
Mail is being reorganized around Go as well. Until now, TechCrunch notes, Mail treated AI as its central tool, and Go is now replacing that role. Users can run enterprise searches across a company knowledge base and connected tools, and Go draws on that context to draft messages and organize inboxes spanning both Google and Outlook accounts.
Twenty new connected agents
Through its Go SDK, Superhuman is adding 20 new connected agents, including DeepL, Legora, Amplitude, Databricks and Blackboard, according to TechCrunch. The move suggests the company wants Go to function as a hub for other agents rather than a single closed assistant.
Meetings feed the context engine
Superhuman acquired the notetaker Fathom in September to bring more context into its suite. Building on that, TechCrunch reports the company is introducing live collaborative notetaking that is shared across a team, and Go can now take action after meetings end — for instance, producing a first draft of a proposal.
One million users and a crowded field
Chief executive Shishir Mehrotra told TechCrunch that Go has passed 1 million active users. He positioned the assistant as work-focused but not limited to work, saying it can still help with personal email, calendars and personal tasks. In his view, assistants of this kind have gained popularity because they address users as complete people covering both professional and personal needs, and he argued Go is simply tilted more toward work than tools emerging from the pure consumer ecosystem.
TechCrunch notes that Go may not compete head-on with the likes of Instinct, Muse and Fo, but Superhuman wants users to choose it over assistants such as OpenAI's Dots or Claude Duet.
Why it matters
Two trends are converging across productivity software: applications are feeding ever more context into their AI features, and they are placing those features on every available surface. Superhuman's rollout is a clear example of both at once. For users, the bet is that an assistant which understands your documents, meetings, inbox and connected tools can do useful work with less prompting. For the industry, the Go SDK and its 20 third-party integrations point toward assistants operating as platforms where other agents plug in, rather than standalone chat windows.
The trade-off is familiar: a context-hungry assistant is only as good as the access users are willing to grant it, and Superhuman has faced criticism of some AI features before, as TechCrunch notes. With more than 1 million active users already, per Mehrotra, the company's combination of deep context and broad surface coverage is now running at a scale that makes it a useful test case for where AI assistants in productivity tools are heading.
- #ai-assistants
- #productivity
- #agents