From On-Demand Chatbots to Always-On AI Desktop Agents
AI desktop agents and ambient AI assistants are moving beyond simple question-and-answer workflows toward continuous, autonomous support. Instead of waiting for users to open a chat window and type detailed prompts, the new generation of workplace automation tools embeds itself directly into daily work. These agents watch how people navigate email, documents, spreadsheets, and internal tools, then begin to automate repetitive actions without needing step-by-step instructions every time. The shift is subtle but profound: AI is no longer just a separate destination or app, but a persistent layer that sits across the operating system and productivity suite. As a result, autonomous workflow agents can anticipate what needs to happen next—drafting responses, organizing information, or scheduling follow-ups—turning knowledge workers’ fragmented tasks into continuous, machine-assisted workflows that run quietly in the background.
Gemini Spark: Google’s 24/7 Workspace Agent
Google’s Gemini Spark exemplifies this move toward proactive workplace automation tools. Announced for Workspace, it acts as a 24/7 personal AI agent that does more than generate text; it takes actions on users’ behalf across Gmail, Calendar, and other apps. Gemini Spark can send emails, add calendar events, and complete long-running tasks in the background while users are offline, effectively functioning as an autonomous workflow agent embedded in the productivity stack. Crucially, Google positions the agent as operating “under your direction,” asking for permission before executing high-stakes actions. This design attempts to balance autonomy with control, reassuring users who may be wary of handing over too much power. By running on Gemini 3.5 and Google’s Antigravity infrastructure, Gemini Spark is built to handle complex, multi-step workflows continuously, shifting AI from a reactive assistant to an always-on coworker inside Workspace.
IrisGo’s Ambient AI Companion on the Desktop
While Google is weaving AI agents into cloud productivity suites, IrisGo is taking a desktop-first approach with an ambient AI assistant that lives on the PC itself. The company’s on-device companion focuses on context awareness and local learning, observing how users work across email, documents, spreadsheets, browsers, and internal tools. Instead of forcing users to translate messy tasks into prompts, IrisGo watches real workflows: drafting emails from documents, pulling figures into reports, summarizing local files, and replaying multi-app routines that usually require multiple tabs and clicks. Operating primarily on Windows PCs via system accessibility features, it can automate actions across traditional interfaces, where many agentic AI systems struggle. Privacy is central to IrisGo’s pitch: by emphasizing local processing and on-device learning, it aims to keep personal files and workflow context on the machine, reframing the assistant as a quiet productivity layer rather than a data-hungry surveillance tool.
Voice-Driven Delegation with Gmail Live and Docs Live
Always-on AI desktop agents are being complemented by more natural ways to tell them what to do. Google is adding voice-driven interfaces to core Workspace apps, turning ambient AI assistants into conversational companions. Gmail Live allows users to search their inbox using natural language—asking questions such as what gate a flight departs from—while the system parses booking emails to extract the answer. Docs Live goes further by transforming spoken thoughts into structured documents, pulling in context from Gmail, Drive, and the web (with permission) to fill gaps or add relevant details. Google Keep is also gaining the ability to turn voice notes into organized lists and notes automatically. Together, these features reduce the friction of task delegation: instead of typing instructions or manually copying content, users can simply talk, letting autonomous workflow agents orchestrate search, summarization, and document creation behind the scenes.
Trust, Distribution, and the Future of Autonomous Coworkers
As AI desktop agents and ambient AI assistants become more proactive, trust and distribution are emerging as decisive factors. A desktop companion like IrisGo, which continuously observes user activity, must be transparent about what it sees, when it acts, and what data leaves the device. Its bet on local processing is both a privacy stance and a product differentiator, supported by seed funding led by Andrew Ng’s AI Fund and a distribution partnership with Acer that could preload the software on AI-ready PCs. Meanwhile, tech giants are rapidly embedding agents into operating systems and productivity suites, with Google pushing Gemini deeper into Workspace and others expanding from chat to full agents. The competitive edge may come from how well these systems learn a few high-value routines, avoid costly mistakes, and integrate seamlessly into daily habits, ultimately defining whether they are perceived as indispensable coworkers or intrusive automation.