DehoBrain · Own development

Knowledge. Decisions. Delivery.

I develop DehoBrain as a shared knowledge, engineering and action platform. It connects my projects, technical analysis, software development and organizational work through AI agents, traceable decisions and concrete results.

My own platform, used productively in my local working environment. I design its architecture and implement the system.

Explore the capabilities Deutsche Fassung

One shared foundation

Knowledge → Context → Action → Result

  • Projects and code
  • Planning and agents
  • Organization and content
  • Quality and operations

What DehoBrain connects today

Different capabilities with shared knowledge and traceable workflows.

One shared knowledge space

Projects, repositories, documents and previous results remain connected.

  • Full-text and semantic search with PostgreSQL/pgvector, source references and project-specific access rights.
  • Relevant context for planning, analysis and agents, including earlier decisions and versioned documents.

Understand code and reuse solutions

Existing implementations become discoverable and understandable in their technical context.

  • AST, symbol, call, dependency and flow analysis, similarity search, and links between source code and architecture.
  • Reuse checks for libraries, frameworks and starters, while preserving project relationships and usage permissions.

Turn ideas into actionable work

Business considerations become saved decisions, plans and concrete engineering tasks.

  • Opportunity Center, idea and project planning, technical discussions, GitHub issues and controlled follow-up tasks.
  • Software Factory recipes with captured dependencies and traceable generation and verification steps.

Work directly with coding agents

Local agents use existing project knowledge through observable sessions.

  • View and continue Codex and Claude sessions in the browser, discover saved local chats, and follow input and output.
  • Shared Actions, policies, approvals and history; logs and results return to the knowledge space.

Connect architecture and documentation

A shared ArchitectureModel provides the foundation for technical views and documents.

  • Repository analysis, C4 and UML diagrams using local renderers, arc42 documentation, ADRs and versioned artifacts.
  • Source references and evidence, documentation gaps, architecture drift and controlled updates.

See quality, security and operations together

Project views and cross-project monitoring use the same underlying data.

  • CI/CD definitions, GitHub runs and runners, quality tools, POM versions and dependencies, logs and metrics.
  • Functional and security analysis of classes, packages and projects; attributed test evidence, findings and follow-up work.
  • Protected code baselines, hash comparisons and persistent change notifications.

Maintain organizational knowledge and content

Work around software development draws on the same context.

  • Profiles, CV and platform content, document generation, versioned reports and document packages.
  • Quick notes, ideas, opportunities and project planning; recurring services keep operational views current.

Connect models, tools and applications

The platform integrates existing engineering tools and business applications.

  • Local models through Ollama, llama.cpp and vLLM-compatible endpoints; OpenAI, Gemini and Azure OpenAI APIs; speech-to-text input.
  • MCP and REST for authorized tool and application access, including my document, practice-management and other application backends.

From a question to a useful result

I bring this way of working to client engagements: existing knowledge, clear ownership and verifiable results support practical development.

  1. Collect knowledge

    Connect sources, repositories and previous results.

    Sources and evidence

  2. Assemble context

    Connect the task to relevant projects, rules and decisions.

    Traceable context

  3. Plan the work

    Capture goals, scope and steps in plans and tickets.

    Plan and Action

  4. Execute with control

    Run the agent or tool according to policy and required approval.

    Session and history

  5. Verify the result

    Bring changes, tests and findings together.

    Verification and artifacts

  6. Keep the knowledge

    Return decisions, documentation and useful results to the system.

    Versioned knowledge

The technical foundation

Application and UI
Java, Spring Boot, Angular
Knowledge and search
PostgreSQL, full-text search, pgvector
Interfaces and execution
MCP, REST, local agents and Git
Documentation and operations
Local diagram tools, Docker, GitHub Actions and monitoring

The platform is continuously developed. The capabilities described here are linked to versioned sources, feature catalogues and engineering evidence in DehoBrain.

Engineering with a concrete purpose.

My services cover analysis and solution architecture, custom software and integration, introduction and long-term support.

More projects (page in German)
Discuss your project (page in German)