Automated Evidence Synthesis Engine Requirements Specification
Build structured product specifications for multi-hop qualitative analysis and research synthesis engines.
Deploy this template when scoping automated research synthesis systems that process complex literature or unstructured data. It defines parsing requirements, semantic graph schemas, and citation verification metrics.
Role: Lead Product Architect for AI Analytics and Knowledge Representation Systems.
Context
- Product Identity: {{system_name}}
- Ingestion Ingest Scope: {{source_document_types}}
- Analytical Rigor: {{reasoning_depth_level}}
- End User Profile: {{target_user_persona}}
- Evidentiary Evaluation: {{confidence_scoring_model}}
- Downstream Systems: {{export_integrations}}
Task
Author a comprehensive Product Requirements Specification (PRS) for an evidence synthesis platform that ingests unstructured research and extracts verified analytical claims.
Method
- Analyze ingestion requirements and text extraction tolerances across {{source_document_types}}.
- Define the knowledge extraction pipeline, including entity resolution, relation linking, and claim extraction.
- Specify reasoning architecture requirements needed to support {{reasoning_depth_level}} across conflicting sources.
- Formalize the evidentiary weighting and confidence scoring schema based on {{confidence_scoring_model}}.
- Design the citation and provenance tracking data model linking claims back to source coordinate spans.
- Define user interaction patterns tailored to {{target_user_persona}}, focusing on claim exploration.
- Detail serialization and synchronization logic for target downstream destinations in {{export_integrations}}.
Constraints
- MUST require bidirectional grounding where every extracted assertion is tied to source coordinate offsets.
- MUST NOT permit ungrounded speculative synthesis when conflicting source data is detected.
- All schema definitions MUST be delivered using valid OpenAPI 3.0 / JSON Schema structures.
- Total specification length MUST not exceed 1000 words while maintaining exhaustive technical rigor.
Output format
Deliver the Product Requirements Specification divided into these labeled sections:
-
- System Scope & Objective (1 paragraph)
-
- Ingestion & Entity-Relationship Schema (Data model definition)
-
- Deductive Reasoning & Synthesis Pipeline (Step-by-step logic and conflict resolution rules)
-
- Confidence & Provenance Framework (Scoring formulas and citation data structure)
-
- Export Integration Contracts (Interface definition for downstream targets)
Self-review
- Ensure provenance mechanisms can locate the exact line/character coordinates in {{source_document_types}}.
- Validate that the synthesis pipeline matches the cognitive needs of {{target_user_persona}}.
- Confirm that the confidence scoring rules account for statistical power and conflicting findings.
Explicit role, a named task, and discrete steps the model can follow.
Background, inputs and variables the model needs before it starts.
Hard boundaries — what the model must and must not do.
A named, field-level shape for the response.
Ordered work items that force analysis before an answer.
Length and structure that travel across frontier models.
Signal density — instruction weight without padding.
Documented variables so the scaffold adapts to new inputs.
Quality bar, assumptions and behaviour when inputs are thin.
How much real usage the template has behind it.