Editorial BrainCortex
Find the signal. Grade the evidence. Publish the truth.
Most research systems are built to store information. Cortex was built to make knowledge usable.
Cortex combines an Obsidian knowledge architecture, connected sources, structured notes, claims, topic maps, retrieval, drafting, and coverage analysis into one living brain for everything I learn.
Raw research enters the system as evidence, becomes connected knowledge, and remains available to question, compare, develop, and publish. Cortex remembers where ideas came from, how they relate, where the evidence is strong, and where the knowledge base is still incomplete.
The result is not simply a vault or an AI assistant. It is a second brain that becomes more useful as its knowledge becomes more structured and connected.
A second brain should not just remember what you know. It should help you build on it.
- Research notes186Sources, claims, concepts, drafts, and topic maps connected inside the current system.
- Indexed knowledge3,783Searchable knowledge units available to the intelligence layer.
- Topic maps9Living maps that connect research, claims, questions, and publishing opportunities.
- Core workflows3Research, drafting, and coverage analysis inside one connected brain.
Screenshots
How Cortex works
It absorbs knowledge
Cortex brings research, sources, notes, quotes, claims, and drafts into one connected system. New information is organized as it enters instead of disappearing into folders or scattered documents.
It remembers with structure
Every item keeps its context: where it came from, what topic it supports, how it relates to other ideas, and where it sits in the publishing workflow.
It connects what you know
Cortex links sources, concepts, claims, topics, and drafts across the knowledge system, revealing relationships that are difficult to see when information lives in isolation.
It reasons across the whole system
Ask a question, explore a topic, or begin a draft. Cortex searches across the knowledge base, combines the strongest relevant material, and returns answers that remain connected to their sources.
It turns knowledge into work
Research can move from raw source to organized note, connected claim, and publishable draft without leaving the system or losing the trail back to the evidence.
What makes it different
One brain, not a collection of tools
Cortex brings the vault, knowledge graph, research workflow, retrieval system, drafting tools, and coverage analysis into one connected environment.
Structure before automation
Sources, notes, claims, topics, and drafts have defined roles and relationships before AI enters the workflow. The intelligence layer reasons over organized knowledge instead of an undifferentiated pile of text.
Knowledge that builds on itself
New research does not disappear after one task. It becomes part of a durable system that can support future questions, comparisons, claims, and drafts.
Gaps become visible
Coverage analysis shows where the system has strong support, where the evidence is thin, and where more research is needed.
From source to publishing
Cortex preserves the connection from raw research through notes, claims, and final work, making it easier to understand how an idea developed and what supports it.
Safe write-back
Generated research and drafts can return to the knowledge system as new files, while create-only rules protect the existing source of truth from being overwritten.
Evidence signals and provenance controls remain available throughout the system, but Cortex is designed to help people think, connect, and create, not simply score sources.
Technical breakdown
Obsidian knowledge architecture
Cortex uses local Markdown and Obsidian as the durable foundation for sources, notes, claims, drafts, and topic maps. Structured frontmatter, controlled vocabularies, folder roles, links, and Dataview indexes turn the vault into a queryable knowledge model rather than a loose collection of documents.
Connected ingestion and knowledge model
New material is classified as it enters the system, enriched with source and evidence metadata, and connected to topics, concepts, claims, and publishing workflows. Claims and quotes can inherit context from the sources they reference, preserving provenance across the system.
Hybrid retrieval and reasoning
The intelligence layer combines semantic search with keyword retrieval, fuses the results, reranks them, and blends relevance with evidence quality. Cortex can then answer questions or develop drafts from the most useful parts of the knowledge system while keeping supporting sources visible.
Coverage, creation, and review
Coverage analytics show where the research base is strong or incomplete. Drafting tools use the vault’s knowledge and brand context, Export Packs support independent review, and create-only write-back allows new work to return to Cortex without overwriting existing knowledge.
Outcomes
- Built a unified second-brain system that moves research from source capture through connected notes, claims, drafting, and review.
- Engineered a hybrid retrieval and reasoning layer that turns a structured Obsidian knowledge base into a queryable intelligence system.
- Created a closed-loop publishing workflow with coverage analysis, auditable source context, independent review packs, and protected write-back.
Lessons learned
- Knowledge becomes more useful when structure survives every stage. Sources, notes, claims, and drafts need clear relationships before an intelligence layer can reason across them well.
- A second brain should compound learning, not just store files. The system becomes more valuable when each new source strengthens future research, connections, and drafts.
- Retrieval quality is a product decision, not a default setting. Search, filtering, reranking, and evidence signals all shape what the system treats as important.
- Automation earns trust when it proposes and connects without overwriting the source of truth. Create-only write-back keeps the human in control of what becomes permanent knowledge.