AI-OS worked example · 13-1111.00

Management Analyst

Management analysis is highly exposed to AI-enabled time compression because the work is rich in data gathering, document review, benchmarking, option development, and recommendation drafting. The danger is not that AI produces too little, but that it produces plausible analysis without enough organizational context. AI-OS therefore treats AI as a research, synthesis, modeling, and first-draft engine while preserving human responsibility for diagnosis, stakeholder judgment, implementation design, and political feasibility.

Method in brief

This public example shows selected outputs from the AI-OS method: task decomposition, permitted AI posture, residual effort, governance burden, capacity compression and role rebundling. It does not expose the underlying diagnostic formulas or implementation workbook.

Decompose

The job title is unpacked into task-level work.

Classify

Tasks are assigned a governed AI posture and reviewed against risk.

Rebundle

The future role is rebuilt around residual human work, proof and accountability.

Time compression

Illustrative task-time estimates, not headcount claims.

120.0h

Baseline monthly work

55.3h

Future residual work

64.7h

Capacity released

AI mode distribution

Distribution by task count.

Automate
27.3%
Assist
54.5%
Advise
18.2%
Human-only
0.0%

Task cluster profile

Data / Analysis / Reporting
40.0h
Finance / Accounting / Budget
36.0h
People / HR / Training
31.1h
Compliance / Legal / Risk
9.9h
Customer / Stakeholder Communication
3.0h

Risk and capability map

High capability, high riskUse AI for preparation, comparison, drafting, and evidence organization. Keep approval, interpretation, and accountability human.
High capability, medium riskGood fit for Assist mode: standardized documentation, summaries, structured analysis, and routine follow-up.
Medium capability, high riskUse only in Advise mode where the work affects people, money, rights, compliance, or institutional trust.
Low capability or high accountabilityKeep human-led where legitimacy, empathy, final judgment, professional assurance, or strategic trade-offs dominate.

Priority task examples

Selected public outputs show how AI mode, governance and control vary across tasks. Detailed scoring rules remain private.

Analyze data gathered and develop solutions or alternative methods of proceeding.

Control: Sample review, audit log, exception queue

Oversight: Exception review

Automate12.2h released
5
Illustrative governance score

Plan study of work problems and procedures, such as organizational change, communications, information flow, integrated production methods, inventory control, or cost analysis.

Control: Sample review, audit log, exception queue

Oversight: Exception review

Automate8.8h released
5
Illustrative governance score

Review forms and reports and confer with management and users about format, distribution, and purpose, identifying problems and improvements.

Control: Sample review, audit log, exception queue

Oversight: Exception review

Automate8.1h released
5
Illustrative governance score

Document findings of study and prepare recommendations for implementation of new systems, procedures, or organizational changes.

Control: Human review before use, source check, version control

Oversight: Review before decision/use

Assist5.8h released
6
Illustrative governance score

Interview personnel and conduct on-site observation to ascertain unit functions, work performed, and methods, equipment, and personnel used.

Control: Human review before use, source check, version control

Oversight: Review before decision/use

Assist5.6h released
5
Illustrative governance score

Rebundled role direction

  • Evidence orchestration and synthesis
  • Human-led diagnosis and change design
  • Implementation governance and stakeholder translation

Control architecture

  • Require source traceability for research and benchmark claims.
  • Separate AI-generated options from human-approved recommendations.
  • Document assumptions, exclusions, and alternative explanations.
  • Validate proposed changes with affected stakeholders before implementation.
  • Treat AI outputs as draft analysis, not organizational diagnosis.

Source note: This example uses O*NET® occupational information from the U.S. Department of Labor, Employment and Training Administration under CC BY 4.0, modified and supplemented for the AI-OS demonstration. USDOL/ETA has not approved, endorsed, or tested these modifications.