Six implemented workflows
Documented applications include lifecycle analysis, business-rule extraction, requirements documents, incident investigation, page mockups, and test paths.
Publication · August 2026
Agentic Workflows for Business Analysts
How structured AI workflows can reduce mechanical analysis work while keeping analysts responsible for scope, validation, judgment, and communication.
Agentic Workflows
for Business Analysts Removing the bottleneck—not the analyst. Steven Duchene · August 2026
for Business Analysts Removing the bottleneck—not the analyst. Steven Duchene · August 2026
01 / THE PREMISE
Remove the bottleneck.
Not the analyst.
Business Analysts create the most value when they resolve ambiguity, evaluate tradeoffs, and help teams make better decisions. Yet much of the role is consumed by researching systems, tracing behavior, assembling documentation, and formatting deliverables.
This paper examines a practical alternative: structured agentic workflows that perform repeatable analysis procedures across code, data, configuration, tickets, and documentation—with human checkpoints at the decisions that require professional judgment.
Verified at the source
Claims about system behavior are checked against code, configuration, ticket history, or database state rather than accepted as model conjecture.
Judgment remains human
Analysts retain responsibility for priorities, tradeoffs, risk, ambiguity, design decisions, and stakeholder communication.
02 / DESIGN PRINCIPLES
Reliability is designed into the workflow.
Hard gates
Critical prerequisites must be satisfied before the agent proceeds.
Interview pattern
Structured questions gather client, scope, and constraint context before execution.
Style calibration
Prior work establishes the expected organizational voice, detail, and conventions.
Layered output
The same analysis can serve client-readable and developer-facing audiences.
Verification loops
Ambiguous findings are checked against authoritative code, data, and configuration.
03 / REPORTED RESULTS
Less production effort.
More analytical capacity.
The paper reports the following effort comparisons from a 10-month enterprise modernization implementation.
| Workflow | Manual effort | Agent-assisted | Reported reduction |
|---|---|---|---|
| Business Rules Extraction | 1–3 days | 2–6 hours | 75% |
| ARD Writing | 60–120 minutes | 5–10 minutes | 90% |
| Incident Investigation | 1–4 hours | 15–30 minutes | 70–80% |
| Page Mockups | 1–2 hours | 10–15 minutes | 85% |
| Test Path Generation | 30 minutes–2 hours | 5–15 minutes | 80% |
Context: These observed ranges describe one implementation—not guaranteed savings or a universal benchmark. Results vary with workflow scope, source quality, tooling, and required human review.
04 / HUMAN ROLE
The analyst moves from executor to director.
The agent handles repeatable research, documentation, and formatting. The analyst scopes the work, resolves ambiguity, validates output, makes judgment calls, and communicates the result.
The workflow establishes what is. The analyst decides what should be.
This division of labor is the design point—not a limitation.
05 / CONTINUE
Interested in the workflow—not just the tool?
Let’s discuss how these principles can apply to requirements, system investigation, structured knowledge, and enterprise modernization.