Cadau, WorkBuddy, and Trae Work: not the same kind of tool

In user language: for team leads choosing workplace AI, or already using WorkBuddy / Trae Work and wanting to see which layer Cadau sits on. All three can “talk to an assistant and get a result”, but what you take away is different: the first two lean toward a person finishing a task; Cadau leans toward a company finishing business work under permissions.

For the user-facing product notes, see [docs/产品介绍.md](/docs/product-intro).

Date: 2026-08-30

The takeaway (read this first)

WorkBuddy and Trae Work are AI office workbenches for people at work: say one sentence, the assistant breaks down the task, calls tools, and hands back a report, slides, a sheet, or code. Cadau is an AI work-assistant platform for enterprises and teams: the assistant remembers the team’s metric, isolates by workspace, and can embed in the systems you already use, or connect HR, warehouse, and compliance onto the workbench.

Do not compare “who is better at slides”. Compare: where the AI appears, where the data lands, and whose work it does.

In one sentence: for a person producing materials, use an office workbench; for a company getting work done by policy, use a team assistant that can grow.

flowchart LR
  U["The thing you need done"] --> Q{"Which layer is this"}
  Q -->|"write a report / make slides / write code / produce a design"| P["Office workbench\nWorkBuddy · Trae Work"]
  P --> A["Take away a deliverable"]
  Q -->|"ask policy / query business numbers / hire-leave-transfer / inbound / run a check or approval"| C["Team assistant platform\nCadau"]
  C --> B["Leave behind a capability that runs under permissions"]

What each one is solving

The first two feel like a very capable personal assistant. Cadau feels like a colleague who clocks in under permissions: workspace boundaries, knowledge metric, authorized business databases. Tell the assistant “help this person onboard”, and it acts within permissions — it does not generate an onboarding slide deck.

Side by side: what customers actually care about

Where they are stronger

Seeing the boundary when you choose is more useful than a hard feature-list comparison.

• Personal output speed: deep research, one-click slides and reports, several experts in parallel — WorkBuddy and Trae Work know this better.
• Coding and design: Trae Work comes from a coding tool; writing code and cloud environments are its home turf; WorkBuddy covers development too.
• Getting started: big-tech accounts, IM entry points, desktop experience — easier for an individual to pick up.

Cadau can also generate office documents and charts, but its home field is not a personal creation tool. If the goal is only “let sales make their own proposal slides”, you do not need Cadau to compete on that layer.

What Cadau is a better fit for

It fits teams that already have business systems, need workspace governance, need to get work done in conversation, and need knowledge plus permissions plus audit. The typical need is not another chat window, but:

• Employees still work in the old HR or OA, and you want the assistant to appear on the page already open, jumping people to the matching feature when needed;
• Policy and project notes must be maintained as a team, answers must have sources, and personal preferences must not be written into public materials;
• Headcount, inventory, and reports must follow grants; the model must not assemble query statements on the fly;
• Hire-leave-transfer, inbound, on-site checks, and approvals must use the same permissions as the page buttons; one sentence to the app assistant should be enough to act;
• The public site and the workbench use the same assistant; if it cannot answer, it hands off to a seat in this workspace.

For front-line members: everyday questions, writing materials, looking up policy; querying numbers and producing charts within grants; letting the assistant remember preferences and project milestones. For team leads: workspaces, inviting colleagues, maintaining team knowledge, turning on assistant capabilities, configuring dedicated agents. For leads who already have a system: embed the assistant, or connect specialist modules to the apps desktop, without replacing the original system first.

They can coexist; you do not have to pick one

Many people already use Tencent or ByteDance office assistants. That does not stop Cadau from coming in.

• A person producing a proposal, writing code, making a design: keep using WorkBuddy or Trae Work.
• Policy Q&A, querying numbers under permissions, hire-leave-transfer, checks and approvals, embedding in existing pages: use Cadau.

Embedding one real page, or opening one workspace as a sample, is easier than a feature list for judging fit. Skill packs can also import and export with common agent-skill formats; it is not a closed spellbook. What an enterprise actually needs to lock down is workspace, knowledge, permissions, business data, and audit — those are governed on Cadau’s own side.

A few questions to ask yourself when choosing

If the other side already has a big-tech enterprise edition or a private deployment: grant that they can also put an office assistant in their own environment. What to compare next is whether the AI enters the business systems you already have, whether it is governed by workspace, whether it can act on hire-leave-transfer, approvals, and checks — not the single sentence “who can privatize better”.

Three sentences to remember at the end

WorkBuddy and Trae Work fit handing “write a report, make slides, write code” to AI. Cadau fits handing “work the company is already running” to AI: isolated by workspace, answered from knowledge, querying numbers and getting work done under permissions, and able to embed in existing systems. When an employee says what they want done, the assistant acts; only when they ask how, do you get a how-to.

So it is not swapping in a tool that chats better. It is giving the team an assistant that can grow, can be governed, and lives on the business.