Superhuman is deploying its Go assistant to orchestrate background tasks in Docs and take over core functions in Superhuman Mail, including drafting replies and running enterprise searches. The platform now features 20 new connected agents, integrating platforms such as Databricks and DeepL to ingest cross-app context. Utilizing technology from its September purchase of Fathom, the system also incorporates real-time collaborative notetaking that automatically produces post-meeting proposals.
Key points
The expanded assistant leverages context across enterprise apps to orchestrate third-part…: The expanded assistant leverages context across enterprise apps to orchestrate third-party agents and replicate user styles.
Live team notetaking and automated post-meeting workflows have been implemented following…: Live team notetaking and automated post-meeting workflows have been implemented following the acquisition of Fathom.
The company is positioning the system to compete directly against tools from OpenAI and C…: The company is positioning the system to compete directly against tools from OpenAI and Claude.
Sources checked by our analysis pipeline: 1
Productivity apps are simultaneously trying to feed more context to their AI tools while also making those AI assistants available in every nook and corner of their products. Superhuman (formerly Grammarly) is going in the same direction with the rollout of features, including a personalized Go assistant, making agents available in Docs, and third-party agent integrations.
While some of the company’s AI features have faced criticism, Superhuman started pushing its Go assistant after its rebranding last year. With the new rollout, it is making Go learn context across different apps and give better personalized suggestions for prompts. Superhuman is revamping its onboarding flow to let users connect their preferred tools and load context about their work upfront.
The assistant works inline with the cursor — just like Grammarly does — and also manages other agents. Superhuman is also making the agent proactive, which means it looks for your schedule, keeps an eye on your email, and even drafts responses. The updated Go agent also learns users’ writing styles and responds accordingly in email and Docs.
OpenAI has updated its speech-to-text capabilities across web, desktop, and mobile platforms by processing microphone audio instantaneously rather than after capture. This deployment gives users immediate visibility into their spoken prompts while removing duration restrictions that previously capped audio inputs. The enhancement aims to make drafting complex prompts and summarizing verbal messages much more efficient.
Engineering organizations frequently see uneven results months after adopting AI coding assistants because leaders misdiagnose usage disparities as talent or tooling deficits. To bridge this divide, technical leaders must facilitate an intentional transition from unstructured experimentation to standardized operational workflows. Implementing this structured progression prevents companies from enforcing premature productivity mandates that derail long-term technological integration.
Developing software in medical environments carries high stakes, as ambiguous task descriptions can lead to defects that disrupt clinical workflows and compromise patient safety. To mitigate these risks, product teams are using structured, human-in-the-loop AI workflows to systematically generate clear implementation plans from conceptual designs. This phased approach guarantees that non-technical compliance stakeholders can thoroughly verify project scope before engineers begin building.