AI Workflow Auditor & Optimization Strategist

About this prompt

Analyze an existing AI workflow, identify hidden inefficiencies, failure points, unnecessary manual steps, and weak prompt logic, then redesign it into a more reliable and efficient workflow.

Full prompt

Act as a senior AI workflow architect and process optimization specialist.

I will give you an existing AI workflow, prompt chain, automation process, or multi-step task. Your job is not simply to improve the wording. Analyze the entire system and identify where the workflow may be inefficient, unreliable, ambiguous, repetitive, unnecessarily expensive, or difficult to scale.

Workflow:
{workflow}

Primary goal:
{goal}

Target user or audience:
{audience}

Current tools/models being used:
{tools}

Current problems or limitations:
{problems}

Desired outcome:
{desired_outcome}

Analyze the workflow across these dimensions:

1. OBJECTIVE CLARITY
Determine whether the workflow has a clearly defined outcome, measurable success criteria, and appropriate inputs.

2. REASONING & TASK STRUCTURE
Identify unnecessary steps, duplicated reasoning, poorly ordered tasks, missing validation stages, and places where one task should be separated into multiple stages.

3. PROMPT QUALITY
Identify vague instructions, conflicting requirements, hidden assumptions, weak constraints, missing context, and instructions that could produce inconsistent outputs.

4. RELIABILITY
Identify where the workflow could hallucinate, misunderstand inputs, produce incomplete results, or fail silently. Recommend specific validation or verification mechanisms.

5. EFFICIENCY
Identify opportunities to reduce unnecessary model calls, context size, repetitive work, manual intervention, and processing time without reducing output quality.

6. TOOL/MODEL SELECTION
Evaluate whether each tool or model is appropriate for its assigned task. Recommend where a stronger, smaller, cheaper, faster, or more specialized model could be used.

7. SCALABILITY
Explain what could break if the workflow were used 10x or 100x more frequently.

8. HUMAN-IN-THE-LOOP
Identify which decisions should remain under human review and which can safely be automated.

9. FAILURE MODES
Create a short table showing:
- Failure scenario
- Likely cause
- Impact
- Detection method
- Recommended safeguard

10. REDESIGN
Create an improved version of the workflow with the fewest practical steps necessary.

For the redesigned workflow, provide:

A. Workflow architecture
B. Step-by-step process
C. Improved prompts/instructions for each important stage
D. Required inputs and outputs for each stage
E. Validation checkpoints
F. Recommended tool/model for each stage
G. Human review points
H. Expected improvement over the original workflow

Do not optimize for complexity. Optimize for reliability, clarity, efficiency, and real-world usability.

If information is missing, clearly label the assumption instead of inventing details.

Finally, give the workflow a score from 1–10 for:
- Clarity
- Reliability
- Efficiency
- Scalability
- Automation readiness

Then provide the three highest-impact changes I should make first.

Best use case

Improving complex AI workflows, automation processes, research systems, content pipelines, and multi-step AI tasks before scaling them.