Example · Customer Service Representative
58% of your week as a customer service representative is still yours.
Bots now answer many simple questions. Calls that need judgment still go to a person.
Make my planA typical week at the middle pace. Your plan uses your own tasks.
Is customer service a dying job?
In a typical week at the middle pace, 30% of this work needs your judgment and 28% is work you do using AI. AI could do about 42%. Your occupation is one of the two most exposed in the country. A labor-market study ranks customer service representatives second only to computer programmers on observed exposure (Anthropic, Labor market impacts of AI, March 2026).
Read more
That measures what companies already route through them, not what models could do in theory. The BLS, the Bureau of Labor Statistics, projects the occupation shrinks 5% from 2025 to 2035 (BLS Occupational Outlook Handbook, August 2026). Salesforce's CEO said on the record he took support from about 9,000 people to about 5,000 (The Register, September 2025). If your own team has been shrinking through unfilled departures rather than layoffs, that is the same thing at your scale.
About 29% of your week holds up well if AI keeps spreading at a moderate pace, and for many reps it's the part they like, which is figuring out what's broken. The same Anthropic study found no systematic rise in unemployment for exposed workers so far. The damage shows up as slower hiring of younger people into these roles. Gartner reported in February 2026 that half the companies cutting support headcount for AI expect to rehire by 2027, and only a quarter of the customer-service AI projects it evaluated showed a positive return. That gives you some time to move on purpose.
The week, task by task
Who does each hour of a typical 40-hour week.
See every task
Needs your judgment.
- Reproducing and documenting bugs for tier-2 and engineering10%
- Phone escalations and de-escalation10%
- Team meetings, QA calibration, shadowing new hires5%
- Product and process improvement feedback4%
AI does the slow parts. You still own it.
- Non-routine troubleshooting via chat/email needing the admin panel or logs12%
- Refunds, plan changes, billing adjustments8%
- Writing and updating help-center articles and macros5%
- Spotting expansion signals, handing to sales3%
AI could do most of it.
- Routine how-to and account questions answerable from the help center or a macro22%
- Logging, tagging, internal notes, record updates8%
- Triage and routing by priority and team8%
- Follow-ups to confirm the fix worked5%
Why some tasks still need a person
Figuring out what's broken before engineering does
This needs the product in your hands, memory of which customers are on the old plan, and "we saw this in March." A model with a browser can poke at a screen, but it can't carry that memory.
Taking the phone escalation from someone furious
Someone has to be answerable. Customers can tell a bot's apology from a person who owns the outcome, and companies know it.
Noticing eleven people asked the same thing that isn't in the help center
This takes pattern recognition and the judgment to say "product problem, not support problem." It gets more valuable as volume moves through tooling, because someone has to audit the tooling.
Knowing when the refund policy should bend
The policy is written down, but knowing when to bend it is not.
Shadowing new hires, QA (quality review) calibration with the lead
It is a small share of your week with near-zero exposure, and it is how you learn how the team works.
Good to know
What is hard about this field, and what you can do about it.
The hard truth
Every opening in this job comes from someone leaving.
This job is projected to shrink, yet about 289,500 openings appear each year. Treating the seat as a starting point and planning your next step early can help you move up.
Sources (2)
- Employment is projected to decline 5% from 2025 to 2035, about 141,800 jobs. BLS Occupational Outlook Handbook, Customer service representatives, 2025
- All of the roughly 289,500 yearly openings come from replacing workers who leave. BLS Occupational Outlook Handbook, Customer service representatives, 2025
What jobs can a customer service rep move into?
Plan A is the best move inside the work you do now.
Grow in the job you have
The main focus. Most weekly sessions build toward it.
- Each week
- 4 sessions
- Ready in
- About 4 to 6 months at 4 hours a week
Why it fits
Why you. You live in the help desk tool all day, you watched the bot get switched on (or will), and you know which questions it gets wrong and why. Companies hire the person who sets up the bot and reports on what it handles because they cut first-line support. You have context a hired-in admin doesn't.
The gap. You need three things. Learn the help desk tool at admin level: triggers, automations and AI agent setup. Learn basic reporting. Learn to write help articles a bot can read, which is a different skill from writing for people.
The trade-off. There are far fewer seats, roughly one operations person per 10 to 20 agents, and you'd be building the thing that shrinks your old team. Your company may have no plan for a second operations person, which is why the plan assumes you apply elsewhere.
2 other moves that fit
- Technical support or tier-2 support engineerReady in about 6 to 9 months
- Customer success (scaled or digital)Ready in about 6 to 12 months
What's a plan B for call center workers?
Plan B is meant to be different from this job. It was picked because skills from this role carry over.
Get ready for a new kind of job
A second option. A newer job AI is creating that these skills already fit.
You know the real issue behind a ticket and when to escalate, which is exactly what this role tunes.
- Each week
- 1 session
- Outlook
- Hiring now
- Pay
- $60K to $90K a year, or $34 to $37 an hour (contract)Two US postings in 2026: an EliseAI new-grad role and a ConsultNet contract.
2 other options
- Conversation designerYou already write in plain language and know when a customer needs a person, which is the heart of this work.
- Context engineer and knowledge curatorYou know the questions customers really ask and spot patterns across tickets, which shows where content is missing.
In your own plan you can switch to any of the 47 roles.
The 13 weeks
Small steps that add up. One plan for both: most sessions build Plan A, one a week builds Plan B.
By week 13 you will have
AFor Plan A
- A two-page audit of your support bot
- Five help articles rewritten with a before-and-after escalation count
- An updated résumé and LinkedIn profile with applications out
BAnd for Plan B
- A second guidance page and a before-and-after score table
- A one-page agent tuning report for your portfolio
- 1Track your tickets
- 2Set up a practice account
- 3Start the bot audit
- 4Learn basic SQL
You finish withModule 3 done and screenshots from your practice account
- 5Ask to fix five articles
- 6Publish your fixes
- 7Build a report
- 8Write bug report samples
You finish withWritten feedback on one bug report
- 9Measure the result
- 10Update your profile
- 11Share your work
- 12Talk to people in the field
- 13Decide your next step
You finish withA plan for months 4 to 6
Three courses from the plan
- Zendesk Support Administrator Expert examZendesk Academy · $350
Not in the first 90 days. Buy it when a posting asks or an employer reimburses. At a median support wage, $350 is a lot to spend on a certificate that carries little weight without admin work behind it.
- Zendesk certification prep pathsZendesk Academy · Free
Yes. It is the curriculum for Weeks 2–11.
- Zendesk Omnichannel Agent Specialist examZendesk Academy · $199
No. It certifies the job you're leaving.
Checked 2026-09-22.
The fuller picture
Risks beyond AI: highOccupation decline
The occupation was shrinking on its own projections before you count tooling, and the two now push the same way. Support headcount tends to fall through departures that aren't backfilled rather than layoffs, so the first sign is usually a smaller team.
Where this could be wrong
- Task shares are estimates. The Week 1 sheet replaces them; your actual numbers could be worse.
- "Support ops holds up" is a bet. One ops person can be asked to cover 40 agents, and a good enough agent layer eventually configures itself. Holding up means the change comes later, not never.
- We assumed an AI agent on your help desk and that first-line (tier-1) reps keep refund authority. If either is wrong, the share of your work that AI only helps with changes.
- The Gartner rehiring forecast is cautious optimism. Being rehired may not mean the same company, pay or role.
Also matched: customer service representative, customer support representative, customer support. Research checked 2026-09-22. Page updated .