Tencent launched a desktop AI agent called WorkBuddy a while ago, and I finally got around to trying it recently.
WorkBuddy isn’t really competing with the products I’d been comparing it to in my head — Cowork, Manus, Cursor. It sits in a different position entirely.
Most of those products are agents. WorkBuddy is something else. A workbench you open to find a roster of agents, each with a name, a specialty, and a usage history. Claude, ChatGPT, Manus, Cursor are individual AIs you talk to. WorkBuddy is a platform where you choose, use, and orchestrate AI specialists. The unit of interaction isn’t “an AI” — it’s “a team of AIs you manage.”
Here’s the clearest way I can put it: many Western desktop agents take their naming from support roles — Copilot, Assistant, Operator, Code Assistant. Tencent named theirs Buddy. And the official copy describes it, plainly, as your dedicated AI colleague. Not assistant. Not copilot. Colleague.
That difference turns out to be the visible tip of a much larger paradigm split.
Western desktop agents usually lead with capability. Their marketing copy and product surfaces consistently describe what the tool can do for you — browse the web, handle your spreadsheets, use your computer.
WorkBuddy leads with relationship and delegation. The framing isn’t “here’s a powerful tool.” It’s “here’s a set of specialists you can delegate work to.”
I noticed this shift first in the product copy. Then I started seeing it in everything else.
The product calls this section the Expert Center. The tagline at the top reads: “Browse experts by industry, invoke them to serve you.” And the first time I opened the Expert Center, it felt less like browsing software features and more like browsing a talent marketplace. Both the teams and the individual experts explicitly listed their specialties, areas of focus, and what they were best at.
The Expert Center splits into two tiers: Expert Teams (multi-role units like “Content Creator Team,” “Deep Research Team,” “Investment Masters Team”) and Individual Experts with proper names — Will, Kai, Phoebe, Sam. They’re filtered by industry the way a staffing platform is: Product Design, Engineering, Finance Investment, Marketing Growth, Sales Commerce, Operations HR. Each card carries specialty tags and a usage count. Will (Senior Developer): 3.15 million uses. Kai (Content Creator): nearly 850,000. You sort by Hottest or Newest. A “My Experts” tab is marked Coming Soon — implying that, eventually, you’ll curate your own roster. The button to engage one of them says “Summon”.
And I tried something open-ended with WorkBuddy — asking it to help me develop a distribution strategy for an essay — it didn’t reply with a draft. It assembled a team. A lead, two specialists (one for domestic platforms, one for international), a scheduling expert. Each one had a name, the way a real colleague would. It set up phases, marked who was working on what, told me two of them were now running in parallel, and would compile the final plan when everyone was done. I was watching a project management dashboard.
Multi-agent systems already exist as a backend pattern. What struck me about WorkBuddy wasn’t the presence of multiple agents under the hood — it was how explicitly the multi-agent structure surfaced in the product itself. The agents weren’t hidden inside a single chat. They had names, profiles, usage counts, and were summoned individually by the user. No mainstream desktop AI product I’ve used surfaces multi-agent collaboration this explicitly as a staffed working structure.
What Tencent shipped is a desktop product built on the assumption that AI agents increasingly occupy positions closer to formal roles inside a working structure, rather than remaining purely tools you operate. The official tagline for this whole approach is unambiguous: a new paradigm for AI agents in the workplace.
Most mainstream AI products still tend to describe AI in terms of assistance, acceleration, or productivity. WorkBuddy feels more comfortable treating agents as delegated roles instead.
A few things this opens up:
Organizational design may become the next bottleneck. If agents start being treated as specialists on your team — named, credentialed, summoned by expertise — existing org charts, headcount logic, and performance structures aren’t built for that model. The question of how these systems integrate into existing organizational logic doesn’t have a clean answer yet — on either side.
The competitive ground may also be shifting from pure model capability to organizational position — where the agent lives in the workflow, how it’s invoked, how trust gets transferred to it, how delegation becomes normalized. That’s a different category of question than model quality, and it may turn out to matter just as much.
A few days into using WorkBuddy, I caught myself doing something I hadn’t done with any other AI agent. I was thinking about how to delegate to it — not how to prompt it, not how to debug it, not how to get more out of it. How to delegate. The verb shift is small, but it’s the kind of shift that, once it happens, is hard to undo.
Whether the rest of the world ends up naming this relationship the way Tencent has is a separate question. But the relationship itself is forming, in different markets at different speeds, under different vocabularies. The most interesting thing about WorkBuddy might not be that it makes AI feel more powerful. It might be that it’s one of the first mainstream products I’ve used that treats AI agents as something closer to organizational roles than software tools.
About the author
Qinqin (Stella) Yang is a director, builder, and imaginist working across AI capability and creative practice. She builds workflows, systems, and creative direction that operate under real constraints, and designs speculative frameworks that expand what teams can imagine before they start making. Over the past decade she has built brand experiences, product design, and speculative projects for Tencent, IKEA SPACE10, Burberry, and the Palace Museum. She is the founder of Actual Fiction, where she develops AI-native tools that turn imagination into a working method.
This newsletter is where she thinks out loud about creativity, AI, and what it means to build toward a future that doesn’t exist yet.
More at qinqinyang.com · Practice at Actual Fiction