Shared core across all libraries
RAG, semantic chunking, HNSW retrieval, MCP servers, agentic workflows, PostgreSQL, Node, and Next.js.
Our Research
A new way of thinking about AI research: methodology-first, not model-first.
Retrieval answers: what evidence exists?
Discursive methodology: how should we think with it - across perspectives - until we can decide?
Our apps belong to a family of methodology-first apps. All share one technical core, and each differentiates through Data Processing Strategies and Discussion Methodologies.
Architecture backbone
From beginning to end, the flow is efficient and reliable.
- 1
Reliable sources are semantically chunked into effective passages.
- 2
RAG retrieval maps every query into vector search over indexed knowledge.
- 3
HNSW graphs accelerate nearest-neighbor retrieval for low-latency context.
- 4
PostgreSQL stores full text, metadata, and citation references for grounding.
- 5
Node and Next.js orchestrate streaming delivery in a production web workflow.
- 6
Specialized agents run discursive methodologies on retrieved context to produce deep insights.
A selected library capable of infinite methodological variations
The platform can be extended with new libraries that share a common core but differ in their data processing strategies and discussion methodologies.
VectorStore-specific implementation
Each library can tune its VectorStore strategy, ranking profile, and corpus scope without losing platform consistency.
Methodology-specific discussion model
Each library adopts a discursive methodology suited to its domain—strategy, law, ethics, product decisions, and more.
Pensador method: Six Thinking Hats
Using the methodology created by Edward de Bono, we deploy six parallel lenses to reduce blind spots and improve decision quality.
White Hat
Facts and evidence
Objective signals, metrics, and known constraints.
Red Hat
Emotions and values
Motivations, fears, and human tension points.
Black Hat
Risk and criticism
Failure modes, weak assumptions, and downside scenarios.
Yellow Hat
Benefits and opportunity
Value creation paths, upside cases, and leverage points.
Green Hat
Creativity and alternatives
Novel options, experiments, and non-obvious reframing.
Blue Hat
Synthesis and governance
Final prioritization, alignment, and actionable direction.
Pensador method: OHP Orchestra (5-stage orchestration)
OHP Orchestra is the execution layer that converts multi-perspective analysis into a defensible decision path, moving from problem framing to implementation guidance.
Stage 1
Decomposition
Split the challenge into decision units, assumptions, dependencies, and constraints so the reasoning scope is explicit.
Stage 2
Interpretation
Interpret retrieved evidence through competing lenses, separating textual support from inference and clarifying conceptual tensions.
Stage 3
Experience
Model lived impact across stakeholders, incentives, emotions, and adoption context to test practical viability.
Stage 4
Audit
Stress-test arguments for logical gaps, bias, unsupported claims, and risk exposure before committing to action.
Stage 5
Synthesis
Integrate validated insights into a coherent recommendation with priorities, trade-offs, and concrete next steps.
Evaluate your thinking with our latest app, exploring the full capability of our library ecosystem.
See how Six Hats and OHP Orchestra perform when your team needs grounded, defensible decisions.