Manus ditches traditional job roles to hire AI-first teams

If you open Manus's career page today, you'll still see typical job listings for frontend and backend engineers. But inside the company, those boundaries have already vanished.

Speaking at Web Summit in Lisbon, Manus co-founder and Chief Product Officer Tao Zhang said the public job board only adheres to traditional titles because "the outside world hasn't caught up yet." Internally, product managers are pushing code to production, and designers are building functional software alongside their mockups.

Ditching LeetCode for a 4-hour prompt audit

The organizational shift started around March. According to Zhang, finding people who simply know how to execute tasks is no longer the bottleneck because AI already excels at raw execution. What Manus needs now are people who know how to collaborate with AI effectively.

That philosophy fundamentally reshaped their technical hiring. Traditional whiteboard algorithms and LeetCode problems have been thrown out the window.

Instead, engineering candidates are given an open-ended practical problem and four hours to solve it using Manus. When time is up, candidates don't just present the final code. They must bring back the complete audit trail of their session, showing every single prompt, clarification, and correction exchanged with the agent.

Interviewers then sit down with the applicant to review the transcript. They evaluate how the candidate framed questions, whether they provided sufficient context, and how quickly they recognized and corrected the AI when it drifted off track. The goal is to see if working with an agent has become muscle memory.

Zhang admitted this mirrors his own daily workflow. Dealing with ADHD, he often found staring at a morning list of 20 to 30 complex tasks overwhelming. Today, he offloads those tasks into Manus, letting the agent pull context across Slack, Notion, and Gmail, delegate subtasks to teammates, and track status updates like an automated Chief of Staff.

Why it matters

While most tech giants treat AI as an autocomplete helper to speed up existing roles, Manus is essentially erasing the walls between disciplines entirely.

For engineers and creators, the takeaway is hard to ignore. Memorizing framework quirks and grinding algorithmic puzzles matters far less today than high-level problem formulation and cross-functional dexterity. When an agent can handle boilerplate syntax, the premium shifts toward product sense, system architecture, and clear human-AI direction.

Still, this hyper-fluid structure isn't something every company can blindly copy. Manus can pull this off because their entire team dogfoods their own autonomous agent all day. In larger enterprises with strict compliance and legacy codebases, letting everyone push code without specialized guards could quickly turn into an unmaintainable mess.

We think Manus is offering a realistic sneak peek into the future of boutique software development. You don't need an army of specialists when a handful of hybrid thinkers can orchestrate autonomous agents to do the heavy lifting.