A practical, task-by-task look · United States source data
AI can draft parts of the work. That is not the same as replacing the whole job.
A systems analyst has to find out what is actually happening in a business, test how a proposed change behaves, and help people decide whether the trade-offs are acceptable. Here’s where AI may help—and where a confident draft still needs a careful analyst.
See the task-by-task breakdownShort answer: the role’s tasks may change as tools change. Available evidence does not let us give this occupation a trustworthy replacement percentage.
Separate a task from a job title
These are practical examples, not measured adoption rates. the O*NET OnLine task list grounds the work described here; the AI analysis is Ikigain’s editorial interpretation, not an occupational exposure score.
01 · FIRST DRAFTS
AI may help: Group meeting notes, suggest a first-pass requirement list, or flag phrases that need clarification.
The analyst still checks: Who was in the room, whose needs are missing, what “fast” or “available” means in this business, and which statements are assumptions rather than agreed requirements.
Related O*NET work: interviewing or surveying workers; analyzing information-processing needs.
02 · TESTING
AI may help: Suggest ordinary, edge and failure cases from a written description.
The analyst still checks: The expected result, real permissions and data, connected systems, exception paths, and whether a test passed in the actual environment—not just in a tidy example.
Related O*NET work: developing and revising test procedures; testing and monitoring systems.
03 · DOCUMENTATION
AI may help: Turn a clear description into a rough flow or summarize long technical notes.
The analyst still checks: What happens when the normal path fails, who owns each step, which records are authoritative, and whether the diagram matches observed work.
Related O*NET work: defining system goals; creating flow charts; documenting design procedures.
04 · INVESTIGATION
AI may help: Sort a supplied error log or generate hypotheses to investigate.
The analyst still checks: The source and time range of the data, what changed recently, whether there is a reproducible fault, and the cost or risk of each proposed fix.
Related O*NET work: troubleshooting malfunctions; reviewing performance indicators; coordinating system changes.
Important: “Human still checks” does not mean those tasks are immune to automation. It means a draft is not yet a verified change, and responsibility, context and risk still have to be handled in the real workplace.
A tiny systems-analysis work sample
A manager wants a fix before the afternoon dispatch. A fictional AI assistant replies:
“This is probably a delayed sync. Retry the sync every five minutes. To keep work moving, track affected orders in a separate spreadsheet.”
Start with evidence: identify a few affected order IDs, when each status changed, the last successful order, and the logs or integration events for those records.
Clarify the process: ask the sales and warehouse teams what each status means and which system is authoritative at each step.
Map the downside: an automatic retry could duplicate an order unless the receiving system is designed to safely ignore repeats; a spreadsheet could create conflicting data and an unowned clean-up job.
Make any test safe: reproduce the issue with test data, define a successful expected outcome, decide who approves the change, and agree how to roll back.
This is one reasonable line of investigation for an invented scenario, not the only correct diagnosis. Real incidents need the system owner, logs and local controls.
Want to try a fuller slice of the work?
Review three fictional order traces, decide what they do—and do not—show, then draft a safe first move. You will leave with a note of your own reasoning, not a score.
Don’t mistake a forecast for an answer about AI
The BLS projects U.S. employment for computer systems analysts to grow 8% from 2025 to 2035 and reports a May 2025 median wage of $105,850. That is useful career context. It does not tell us how much AI caused the projection, whether every region will grow, or whether an individual job is safe.
The ILO’s 2025 global study estimates that one in four workers are in occupations with some GenAI exposure, while concluding that transformation is more likely than full job redundancy for most jobs. That is broad global context, not a prediction for U.S. systems analysts—and it does not cover every form of automation.
Our best honest answer is therefore specific but not falsely precise: AI can assist with some information and drafting tasks; the job also includes understanding real processes, checking systems, coordinating changes and making decisions in context. How that balance changes depends on the employer, tools, industry and task mix.
See what each source supports ↓Would you want to do the real work?
Maybe you liked finding the missing evidence. Maybe the stakeholder questions sound draining. Both are more useful clues than a made-up “AI-proof” score. Save your reaction, the concern you’d investigate, and one small next step.
Save a thought and next step ↓Make this useful to you
Not a score or a career verdict. Just your reasons to look closer—or move on.
Notes stay in this browser only. Avoid sensitive personal information. Saving notes also adds this role to your saved paths.
Sources and method
Role definition and reported tasks, including user support, system testing, requirements analysis, workflow changes and user interviews.
U.S. wage and employment context. Its projections describe employment, not the causal effect of AI on this role.
Global task-level GenAI exposure context. It is not a systems-analyst-specific replacement forecast.
Sources checked 28 September 2026. U.S. figures are not local guidance for Latvia or other countries. The task-by-task AI commentary is Ikigain’s editorial analysis, not a measured adoption study, expert-reviewed risk assessment, validated career test or guarantee.
Occupational information on this page is adapted from O*NET OnLine, sponsored by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA), under the CC BY 4.0 license. Ikigain has adapted and interpreted the information; USDOL/ETA has not approved, endorsed or tested these modifications. O*NET® is a trademark of USDOL/ETA.
Saved paths and work-sample notes stay in this browser. Fictional exercises are for exploration, not professional assessment.