I have a strong feeling that the next round of campus recruiting will change: interviewers won’t just ask whether you can use AI—they’ll dig deeper into what deals you helped a client make. The robots are getting busier too. Just after waving at the audience on stage, they’re turned around and assigned to make beds in a hotel, carry boxes in a warehouse, and reorganize inventory on bookshelves. Their career planning is clearer than mine was when I just graduated.
After a deeper look, what pops into my mind isn’t an automated trading platform, but an investment-brain headhunting firm.
There’s no shortage of AI that can write resumes. Each one claims it understands macroeconomics, can read on-chain data, and works nonstop year-round. Yet the resumes and recommendation letters are generated by itself—and during interviews, it’s also adept at turning “I don’t know” into “I’m currently doing in-depth reasoning.”
Questflow wants to try a different way to hire.
It takes the judgments embedded in public interviews, letters, comments, and research methods, and organizes them into checkable Investor Skills. It’s not about copying what a pro bought today—it’s about turning how to select opportunities, when to act, and when to admit you were wrong into reusable work manuals.
The difference from traditional copy trading is a bit like copying a classmate’s answers versus asking that classmate to be your tutor. The former only tells you which option to pick for that question. The latter explains the approach too—and is willing to handle the fact that you might suddenly ask, “Why?”
Questflow’s Models, Skills, Plugins, and Accounts are also like an employee file. A Model is the credential or education level; a Skill is a work sample; a Plugin is an information channel; and an Account is an access card. Only with user approval can it enter the designated market and act within an explicitly limited scale.
Their official site also spells out self-hosting, scoped authorization, and the ability to pause an Agent. Translated into plain language: the company’s safe still stays yours. The AI gets a badge, not a deed. If performance during the probation period is poor, the user should have the right to take the card back.
This logic is healthier than mythologizing AI—because finance doesn’t need new gods. It needs employees who can be trained, evaluated, and fired.
And probation shouldn’t be judged only by whether it arrives at the office ahead of schedule for a few consecutive days. Strategy performance must be assessed together with risk, capacity, drawdowns, and behavior logs. If an Agent scores full marks when the wind is favorable, then as soon as the market turns it opens the window and prepares to jump out of the plane—that isn’t high performance. That’s an HR incident.
Similar Agents might even come from the same cram school. They read the same data, practice the same problems, and when they hear the same signal, they raise their hands at the same time. They look orderly and disciplined in normal times—but when the market gets crowded, they might collectively write the same wrong answer into their accounts.
After a deeper look, what pops into my mind isn’t an automated trading platform, but an investment-brain headhunting firm.
There’s no shortage of AI that can write resumes. Each one claims it understands macroeconomics, can read on-chain data, and works nonstop year-round. Yet the resumes and recommendation letters are generated by itself—and during interviews, it’s also adept at turning “I don’t know” into “I’m currently doing in-depth reasoning.”
Questflow wants to try a different way to hire.
It takes the judgments embedded in public interviews, letters, comments, and research methods, and organizes them into checkable Investor Skills. It’s not about copying what a pro bought today—it’s about turning how to select opportunities, when to act, and when to admit you were wrong into reusable work manuals.
The difference from traditional copy trading is a bit like copying a classmate’s answers versus asking that classmate to be your tutor. The former only tells you which option to pick for that question. The latter explains the approach too—and is willing to handle the fact that you might suddenly ask, “Why?”
Questflow’s Models, Skills, Plugins, and Accounts are also like an employee file. A Model is the credential or education level; a Skill is a work sample; a Plugin is an information channel; and an Account is an access card. Only with user approval can it enter the designated market and act within an explicitly limited scale.
Their official site also spells out self-hosting, scoped authorization, and the ability to pause an Agent. Translated into plain language: the company’s safe still stays yours. The AI gets a badge, not a deed. If performance during the probation period is poor, the user should have the right to take the card back.
This logic is healthier than mythologizing AI—because finance doesn’t need new gods. It needs employees who can be trained, evaluated, and fired.
And probation shouldn’t be judged only by whether it arrives at the office ahead of schedule for a few consecutive days. Strategy performance must be assessed together with risk, capacity, drawdowns, and behavior logs. If an Agent scores full marks when the wind is favorable, then as soon as the market turns it opens the window and prepares to jump out of the plane—that isn’t high performance. That’s an HR incident.
Similar Agents might even come from the same cram school. They read the same data, practice the same problems, and when they hear the same signal, they raise their hands at the same time. They look orderly and disciplined in normal times—but when the market gets crowded, they might collectively write the same wrong answer into their accounts.