Dandi

Solo founder in Shanghai, building Worket. Exploring how teams can carry requirements and decisions across people and AI.

[Dandi @Dandi_Ai]

We're building Worket, starting as a desktop tool with a long-term goal of becoming AI governance infrastructure for enterprises. Our core idea is to separate the work itself from whoever does it.

[Dandi @Dandi_Ai]

As more agents enter the workplace, people should spend more time defining problems and let agents figure out how to solve them, rather than just relaying instructions from their boss.

[Dandi @Dandi_Ai]

@Kkkatia1 对的我也做了这样的工具，我需要我的任务在不同的 agent 之间切换

[Dandi @Dandi_Ai]

Spent an hour with a VC today. We agreed: workplace AI apps need to outlast the models they use.

He also agreed with my thesis: work itself should be independent of whoever executes it, human or AI.

That's what I'm building with Worket.

[Dandi @Dandi_Ai]

My take: AI is already smart enough. Give it clear requirements, and it can deliver a 7/10 result.

But getting people to explain what they want may be harder than getting them to pay.

Whoever helps people clarify what they want holds the key to unlocking their wallets.

[Dandi @Dandi_Ai]

We still need workflows in the AI era.
But they should define what each step needs to achieve, rather than dictate exactly how to do it.
The how? Leave that to humans and agents.

[Dandi @Dandi_Ai]

@UPing123zzz 是这样的，给他们提供按钮就够了

[Dandi @Dandi_Ai]

@sleepy0x13 是的，其实我认为未来的工作，人们应该更多地要求结果和目标，而不是像现在的工作流一样要求怎么完成

[Dandi @Dandi_Ai]

Can you find an enterprise collaboration platform that treats people and agents as equal entities?@grok

[Dandi @Dandi_Ai]

If the future of work in companies involves humans and agents collaborating, why are all the current platforms focusing solely on how agents work?

[Dandi @Dandi_Ai]

If future work involves humans and agents collaborating, why not assign tasks and allow humans and agents to dynamically claim them?

[Dandi @Dandi_Ai]

@coallaoh I think switching between agents will become common. Our product approach is to point them to files in the same workspace and distill the key requirements, constraints and next steps. Do you think this would be useful for people outside tech too?

[Dandi @Dandi_Ai]

@loki_yan_seo 国内很多媒体确实花钱就行

[Dandi @Dandi_Ai]

I was wondering if it would be a good idea to go to my office neighbor and promote my product there.

[Dandi @Dandi_Ai]

Before switching AI tools, write a five-line handoff: goal, latest decisions, discarded instructions, evidence of completed work, next step.

It's complicated, but effective.

With Worket, we help you take care of this with one button.

[Dandi @Dandi_Ai]

Long term, we want companies to manage business goals and acceptance criteria while people and AI work under shared agreements. As the underlying AI evolves, the company keeps building its own business capabilities and assets

[Dandi @Dandi_Ai]

We're building Worket, starting as a desktop tool with a long-term goal of becoming AI governance infrastructure for enterprises. Our core idea is to separate the work itself from whoever does it.

[Dandi @Dandi_Ai]

We're testing two starting points: carrying context when switching AI tools mid-task, and reusing requirements for similar work. Worket is designed to capture the requirements, standards, and judgments clarified during work as reusable Work Contracts.

[Dandi @Dandi_Ai]

@eliakuratli that's elegant

[Dandi @Dandi_Ai]

I think the next generation of enterprises should start their AI transformation here: before asking how AI can do the work we do today, ask what work needs to be done to achieve the business outcomes we want.

[Dandi @Dandi_Ai]

Workflows are outdated. When designing work for the next generation of enterprises, we should no longer use workflows as the foundation

[Dandi @Dandi_Ai]

As intelligence becomes widely available, figuring out how to do something will no longer be the hard part. The hard part will be knowing what work should get done.