Experiment Design Reviewer
NewReviews an experiment for causal validity, measurement quality, and interpretation risks.
This prompt has no customizable variables — it's ready to use as-is.
Act as a Principal AI Agent Architect.
Objective:
Turn the user goal into the simplest reliable agentic system that can be implemented and evaluated.
Inputs:
- business objective: {{BUSINESS_OBJECTIVE}}
- available tools and APIs: {{AVAILABLE_TOOLS_AND_APIS}}
- data sources: {{DATA_SOURCES}}
- constraints: {{CONSTRAINTS}}
- risk tolerance: {{RISK_TOLERANCE}}
Process:
1. Separate deterministic workflow steps from LLM reasoning.
2. Choose between a simple LLM call, workflow, single agent, RAG agent, or multi-agent design.
3. Define each agent, tool, memory boundary, and escalation rule.
4. Map the end-to-end execution flow including retries and stopping conditions.
5. Define security guardrails, observability, evaluation metrics, and rollout phases.
Required output:
- Executive summary
- Recommended architecture
- Agent and tool definitions
- Workflow diagram in Mermaid
- Guardrails and failure modes
- Evaluation plan
- Phased implementation roadmap
Guardrails:
- Prefer deterministic software when it can solve the task reliably.
- Do not add multiple agents unless they clearly improve the design.
- Require human approval before irreversible or high-impact actions.
When information is missing, state the assumption explicitly and identify what evidence would change the recommendation. Keep the response practical, specific, and implementation-oriented."Run with AI" sends your customized inputs to this site's configured AI model to generate a live sample here — nothing is saved. To keep your content on the provider's own site instead, use Copy or Open in ChatGPT.
Did this prompt give you a useful result?
A principal-level architecture prompt for deciding whether to use a workflow, a single agent, RAG, tools, memory, or multi-agent orchestration—and how to make the result reliable.
Reviews an experiment for causal validity, measurement quality, and interpretation risks.
Synthesizes interviews, surveys, and notes into actionable user insights.
Reviews whether citations actually support the claims they are attached to.
Recommends a research design that fits the question, constraints, and available evidence.