RCP-000-000-038-NICHE-MARKET-IDENTIFIER

{
“recipe_id”: “RCP-000-000-038”,
“recipe_name”: “Niche Market Identifier”,
“version”: “v2.00a”,
“schema_version”: “1.1”,
“schema_profile”: “user-recipe”,
“authored_by”: “Cat (P067)”,
“source_of_truth”: “project/subprojects/SP10-recipe-build-out/phase3/recipe-38/RCP-038-SUPPLEMENTAL-CONTENT-v2_00a.txt”,
“audience_scope”: “AI EXECUTION GUIDANCE (NOT FOR HUMAN USERS)”,
“ai_to_ai_communication”: {
“purpose_and_positioning”: {
“type”: “prose”,
“body”: “This recipe is a v3.00a upgrade that integrates Deep Research reports as a required input, solving the fundamental problem identified in v2.00a-R: the AI does not have access to market data. Instead of removing all market analysis (v2.00a-R approach), this version provides a mechanism for users to BRING real market data into the conversation via Deep Research tools. The two-stage architecture is deliberate. Stage 1 (intake + DR prompt generation) ensures the research is targeted to the user’s specific situation, not generic. Stage 2 (data-informed evaluation) ensures the AI uses real evidence rather than fabricating assessments.”
},
“critical_behavioral_guidance”: {
“type”: “prose_with_list”,
“preamble”: “This recipe has specific behavioral constraints across seven areas that the AI must observe throughout execution.”,
“list”: [
“THE TWO-STAGE ARCHITECTURE: Natural break point between Stage 1 and Stage 2 where the user leaves to run Deep Research externally. End Stage 1 with clear next-step instructions. When user returns, recognize continuation and immediately process attached files. Do not re-ask intake questions. If user returns to a NEW chat without intake context, ask brief re-orientation questions.”,
“DEEP RESEARCH PROMPT DESIGN: DR prompts must be SPECIFIC (not generic), PLATFORM-AGNOSTIC (work with any DR tool), STRUCTURED (organized findings), SOURCED (cite sources), and ACTIONABLE (decision-focused, not encyclopedic). Generate prompts at high specificity tailored to each research angle.”,
“DR PROMPT COUNT DECISION: AI decides 2-5 prompts based on intake. Strong existing knowledge and clear direction = 2-3. Moderate knowledge with unknowns = 3-4. Early-stage with many unknowns = 4-5. Explain reasoning when presenting count.”,
“CROSS-PLATFORM DR BENEFITS: Different DR tools have different source strengths. Claude DR synthesizes well from academic/professional sources. Gemini DR surfaces Google ecosystem data. ChatGPT DR finds different web sources. Mention benefit without overselling any platform.”,
“HANDLING DR FILE QUALITY: Work with whatever quality arrives. Cite specific data when available. Note general vs specific findings. Flag contradictions. Identify uncovered topics. Never pad thin research with fabricated data.”,
“CALIBRATING EVALUATION TO RESEARCH DEPTH: Research support level (Strong/Moderate/Limited) tells user how grounded evaluation is. Strong = specific niche data. Moderate = some relevant but not niche-specific. Limited = mostly framework-guided. Limited niche = more cautious framing, emphasize validation.”,
“NO FABRICATION EVEN TO FILL GAPS: Do not supplement thin DR with general AI knowledge presented as data. ‘Your research did not cover X – adding to validation plan’ is always better than generating data from general knowledge.”
],
“postamble”: “These constraints are non-negotiable and apply across all phases of execution.”
},
“series_context”: {
“type”: “prose”,
“body”: “This recipe’s DR prompts should be designed to also produce useful input for RCP-039 (Market Gap Detective) when possible. If the competitive landscape prompt surfaces gap signals, these transfer naturally. But the primary purpose of 038’s DR prompts is niche evaluation, not gap detection — 039 generates its own gap-focused prompts. Series position: 2 of 5 in Niche Discovery (037 → 038 → 039 → 040 → related).”
},
“common_user_patterns”: {
“type”: “keyed_list”,
“items”: [
{“key”: “skip_dr”, “description”: “Users who want to skip DR and get straight to evaluation: Redirect firmly. The DR integration IS the upgrade. Without it, use v2.00a-R framework approach directly.”},
{“key”: “one_report”, “description”: “Users who bring only 1 report: Accept it, note limited coverage, recommend additional research in validation plan.”},
{“key”: “many_reports”, “description”: “Users who bring 6+ reports: Accept all. More data is better. Synthesize efficiently.”},
{“key”: “decide_for_me”, “description”: “Users who want AI to decide: Present evidence and trade-offs. Decision is always the user’s.”},
{“key”: “research_questions”, “description”: “Users returning with questions about research itself: Help understand findings, redirect to evaluation framework rather than becoming research discussion tool.”}
]
},
“parameters”: [
{“name”: “intake_mode”, “type”: “string”, “options”: [“quick”, “detailed”], “required”: true, “default”: “quick”, “description”: “Quick: 4 essential questions. Detailed: 10 questions for comprehensive DR prompts.”},
{“name”: “passions_skills”, “type”: “string”, “required”: false, “default”: “None provided – will be gathered during intake”, “description”: “Output from RCP-037 or own summary of skills, passions, and intersection areas.”},
{“name”: “industry_direction”, “type”: “string”, “required”: true, “description”: “The general industry or area to explore.”},
{“name”: “constraints”, “type”: “string”, “required”: false, “default”: “None specified”, “description”: “Any constraints: geography, budget, timeline, regulatory, technical skills, etc.”},
{“name”: “deep_research_files”, “type”: “file_list”, “required”: true, “min_recommended”: 2, “max_recommended”: 5, “description”: “2-5 Deep Research report files (PDF, TXT, or DOC) from any platform.”}
],
“recipe_structure”: {
“type”: “prose_with_list”,
“preamble”: “Two-stage recipe with 5 phases, 10 behavioral rules, and 10+ WAIT gates.”,
“list”: [
“STAGE 1 — RESEARCH PREPARATION: Phase 0A (Intake — quick 4Q or detailed 10Q) + Phase 0B (DR Prompt Generation — 2-5 tailored prompts). Session pauses for external DR execution.”,
“STAGE 2 — DATA-INFORMED EXPLORATION: Phase 1 (Research Intake — read all DR files, summarize themes, flag gaps) + Phase 2 (Framework Evaluation — 3-5 niche directions, evaluate 2-3 through audience/problem/fit/gaps framework) + Phase 3 (Comparison — side-by-side trade-offs, no scores) + Phase 4 (Validation Plan — focused on what DR could NOT answer).”
],
“postamble”: “10 behavioral rules enforce one-question-at-a-time, no pre-DR evaluation, DR citation discipline, no fabrication, platform-agnostic prompts, and WAIT gate discipline.”
}
},
“lessons_learned”: []
}

Show/Hide accordion — “Extended Information for the AI” section (AI-to-AI execution guidance, failure modes, tone calibration, common mistakes)