Three-Bucket Career Task Audit
Decompose your job into tasks, then classify each by AI exposure
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 98%
Stop treating a job title as the unit of career analysis. Instead, describe the 10 to 20 tasks that make up the actual working day and sort them into three buckets: highly automatable, materially improved by AI, and unlikely to be automated because they depend on human judgment or interaction. The resulting distribution exposes the role's real risk and opportunity. A role dominated by automatable tasks is a prompt to investigate alternatives; a mixed role calls for adopting AI in the assisted bucket while developing the human capabilities that protect the third. Repeating the audit as tools evolve turns a vague fear about AI into a practical career-development agenda.
Origin
Ryan Roslansky and Anish Raman present the three-bucket task framework in Open to Work as a way to understand how AI changes jobs.
Core principles
- 01A job title hides the work that actually creates value
- 02Different tasks within one role face different levels of automation
- 03AI can augment a task without fully replacing it
- 04Career decisions should follow the task mix, not the title
How to run it
- 1
Remove the title
Describe what you do without using your job title. Focus on observable work rather than the label attached to it.
Pro tip Review a typical week so recurring but easily forgotten tasks are included.
Watch out Do not assume the title accurately represents your contribution.
- 2
Inventory the tasks
Write down the 10 to 20 tasks that matter most in your day-to-day work. Keep each item specific enough to assess independently.
Pro tip Use verbs such as draft, decide, reconcile, coach, or negotiate.
Watch out Broad entries such as 'manage projects' conceal several different tasks.
- 3
Sort into three buckets
Classify every task as highly automatable, AI-assisted, or unlikely to be automated. Judge the task's mechanism, not whether you currently use AI for it.
Pro tip Test uncertain tasks with a real AI tool before classifying them.
Watch out Do not confuse AI assistance with complete automation.
- 4
Read the task mix
Estimate which bucket contains most of the role's value and time. Treat a role dominated by automatable tasks as a material career signal.
Pro tip Weight mission-critical tasks more heavily than administrative volume.
Watch out A simple count can mislead when one high-value task matters more than ten minor ones.
- 5
Build the adaptation plan
Automate bucket-one work, learn to use AI on bucket-two work, and deliberately practise the human skills behind bucket three. Investigate adjacent roles if the first bucket dominates.
Pro tip Re-run the audit as tools and responsibilities change.
Watch out Do not wait for the employer to design the transition for you.
In the wild
A product manager separates customer research, prioritisation, interface drafting, coding, stakeholder persuasion, and conflict resolution. AI assists interface and code production, while judgment and alignment remain human. The manager uses the tools to build prototypes directly and invests more practice in customer conversations and trade-off decisions.
→ The role becomes broader and faster rather than simply disappearing.
Common mistakes
Auditing the title instead of the work
A title-level judgment treats a mixed role as uniformly safe or exposed. The framework only works when the role is decomposed into specific tasks.
Treating assistance as replacement
A task that AI accelerates may still require human direction, review, or accountability. Keep augmentation distinct from full automation.
Is it for you?
Best for
Professionals who need a concrete way to assess how AI will change their current role.
Not ideal for
People seeking a precise prediction of when a particular employer will eliminate a role.
From the transcript
“Again, everybody's job is is a set of tasks.”
“And Anish and I in the book talk about it in three buckets. Like, bucket one, highly automatable tasks. Uh bucket two, the types of…”
“Sit down this week and list your [music] top 10 to 20 tasks and weekly responsibilities, and start sorting them honestly into three buckets.”
From the episode
Ryan Roslansky: Former LinkedIn CEO on Building an Irreplaceable Career in the Age of AI
Ryan Roslansky