sunjian

Wery is an AI Expert Workspace. Tell Wery the goal. It coordinates the right AI experts, previews the execution plan, and runs tasks in parallel to get real work done. Not another chatbot. It's a real work system.

I’m building Wery, an AI Expert Workspace that acts as your virtual studio. Instead of a generic bot trying to do it all, Wery is a specialized system built to orchestrate and execute complex tasks. So, what are your biggest pain points? I’d love to hear what you’re working through.

Key Insights Today : 1. Until true AGI arrives, expert agents are here to stay for the long haul. 2. The road to AGI will likely depend on a massive scale of expert agents reaching critical mass, sparking an emergent qualitative leap. 3. Expert agents are the ultimate vessels for human intelligence—which is exactly what the next phase of large model training demands. ----- 1、在真正的 AGI 出现之前，专家 Agent 必然长期存在 2、实现AGI 的路上，可能需要依赖海量专家 Agent 从量变到质变的涌现 3、专家 Agent 是最好的人类智慧载体，这正是大模型训练的下半场所需要的

@DSmith_Tucson an original take

@sunjian AI is a pestilence.

One thing feels increasingly obvious to me: AI is not moving toward better chat. It’s moving toward better work orchestration. If you’re shipping something real, the hard part usually isn’t getting one decent answer. It’s keeping research, writing, slides, visuals, and video from breaking apart between steps. That’s the bottleneck I care about most: less switching less waiting less context reconstruction The next useful AI products won’t just generate. They’ll organize the run. Where does the handoff still break for you?

High-value work shouldn’t start with “trust me.” It should start with a visible plan. Not because plans look sophisticated, but because users deserve scope, intent, and control before the system starts burning time, budget, and context. When users say an AI product feels expensive, I usually hear one of three things: I couldn’t predict the spend. I couldn’t control the run. I didn’t trust the result enough to build on it. That’s not just pricing. That’s product design.

The deeper I get into this category (AI Expert Workspace), the more I think there are two kinds of memory that matter: project memory and method memory What are we working on? And how do I like this kind of work done? Most systems only solve half of that.

Every AI product now seems to want the same nouns: agent, coworker, workspace, computer, team. The label isn’t the interesting part anymore. The real question is: after the first run, is the work actually more organized, more controllable, and more reusable?

Most AI products are still built around answers. Real work needs something else: one accountable entry point specialists when needed a plan before execution parallel progress outputs that stay usable That’s the problem we’re building Wery around.