daBongo LMS AI Training Courses

← Back to Course

Claude Cowork in Practice – Real Files, Real Workflows

Lesson 5: Steering Multi-Step Work Responsibly

Lesson Objectives

By the end of this lesson, students should be able to:

  • Set appropriate autonomy levels for different types of Cowork tasks
  • Identify early signals that a multi-step task is drifting from the goal
  • Use steering input to correct course without restarting
  • Apply heightened oversight for tasks with real-world consequences

Lesson Content

Calibrating autonomy to task stakes.

Not all Cowork tasks warrant the same level of oversight. The right autonomy level matches the reversibility and consequence of the task:

Low-stakes, reversible tasks: Higher autonomy is appropriate. Claude can work through multiple steps with minimal check-ins. Example: synthesizing documents, drafting content, summarizing research. If something goes wrong, it is caught in the final review and easily corrected.

Medium-stakes tasks: Checkpoint-based oversight is appropriate. Define specific points in the task where Claude should confirm before proceeding. Example: large document productions, multi-step analysis. Check in at logical phase boundaries.

High-stakes, hard-to-reverse tasks: Step-by-step review is appropriate. Claude should confirm before every significant action. Example: any task involving write access to connected apps, production communications, financial calculations.

Early signals a task is drifting.

In multi-step work, errors compound – a wrong early step can invalidate later work. Watch for:

  • Output that addresses a different question than the one you asked
  • Scope creep: Claude is doing more than assigned, touching things you did not ask it to touch
  • Format that does not match your instructions
  • Confident claims about factual matters that seem inconsistent with your knowledge
  • Tools or resources being used that you did not specify

Using steering input to correct course.

When you notice drift, specific steering is more effective than a generic "that's wrong":

  • "You are summarizing the wrong document – I need the 2026 version, not 2025."
  • "Your scope has expanded into section 4 of the report – I only need sections 1 and 2."
  • "The format here is wrong – I need a table, not a narrative."

Specific steering preserves the work done so far and redirects only the errant element.

Heightened oversight for consequential tasks.

For any Cowork task that connects to real apps, produces output for external distribution, or involves data that affects real decisions: use Claude in a "show me before you do" mode. In the task assignment, include: "Before executing any action in [connected app], describe what you are about to do and wait for my confirmation." This adds a review step before each consequential action.

Practical Example

A operations coordinator runs a Cowork task to draft emails to twelve vendors from a template, using data from an uploaded spreadsheet.

She sets the task at "show me before you send" mode – Claude drafts each email and presents it to her before executing the send action through the connected email plugin.

On the third email, Claude has populated the wrong vendor name from the spreadsheet (a row alignment issue in the data).

She catches it before it sends.

The heightened oversight mode costs five extra minutes of review across twelve emails – worth it to catch a data error that would have created confusion with a vendor.

Lesser-Known Tip

After completing a multi-step Cowork task, ask Claude to provide a brief task log: "Summarize what actions you took during this task, what sources you used, and any decision points where you made a judgment call without asking me." This audit trail helps you evaluate whether the task ran as intended and identify where your assignment instructions could be clearer for next time.

Safety Notes

Multi-step autonomous work in Cowork is a powerful capability – it is also one where errors can propagate silently through multiple steps before surfacing. For any task with real-world consequences – external communications, data modifications, production content – set explicit check-in requirements in your assignment and review final outputs carefully before use. The review discipline that protects you in manual work applies equally when the work is automated.

Practice Task

Identify a multi-step Cowork task from your work and assign it an autonomy level: high, medium, or low, based on reversibility and real-world consequence. Write the appropriate oversight instructions for that autonomy level and add them to a task assignment. If you can run the task, do so and apply your oversight instructions – note whether the check-in points you defined were the right ones.

Completion Check

You should be able to assign an autonomy level to different types of Cowork tasks, identify three early signals that a task is drifting, write specific steering input to correct course, and know when to use "show me before you do" mode.

Log in and enroll to access lesson quizzes.

← Back to Course

Scroll to Top