Wissensbau
DE Sign in

Non-commercial research project

How do scattered documents become a reliable knowledge build?

Wissensbau studies how language models can condense raw notes, emails and documents into a linked wiki – following Andrej Karpathy’s LLM-wiki pattern. The prototype is the testbed for our experiments.

PDF · DOCX · XLSX · PPTX · Email · CSV · Images · Audio

4
pipeline steps from raw text to wiki
6
research questions in focus
2
interface languages: DE and EN
0
trackers, cookies only for sign-in

wiki / hardware / battery

„How long does the sensor battery last in continuous use?“

Principles

What guides the project

1

Verifiable over persuasive

Every statement in the wiki should trace back to its source. An answer without evidence counts as a bug, not a feature.

2

Open and non-commercial

Wissensbau is a research and learning project with no product promise. There are no ads, no subscriptions and no user profiles.

3

Frugal with data

Self-hosted, with no trackers and no external fonts. Personal data can be redacted before it is processed.

Method

The pipeline in four steps

  1. 1

    Collect

    Raw data stays untouched in a folder – files, emails, or connected sources such as IMAP and PostgreSQL.

  2. 2

    Compile

    A language model extracts topics, entities and concepts and writes wiki pages from them.

  3. 3

    Link

    Pages are cross-linked; relationships show up in the knowledge graph.

  4. 4

    Query

    A chat answers from the wiki and names the sources behind each answer.

Research questions

What the prototype investigates

Extraction across formats

How reliably can text, email, PDF, Office documents, spreadsheets, images and audio be turned into structured knowledge?

Synthesis and linking

How well does raw material become consistent, cross-linked pages that point back to their sources?

Grounded answers

Do chat answers demonstrably rest on the wiki? A faithfulness evaluation checks this.

Entity resolution

Are people, projects and concepts merged correctly across documents and reflected in the graph?

Hybrid retrieval

Keyword search versus embeddings – including a benchmark of how each scales with corpus size.

Privacy and trust

Redaction of personal data and source trust as building blocks for responsible use.

Application areas

Where the pattern can be tested

Not a finished product, but scenarios in which we try the method and probe its limits.

  • Research groups

    Condense lab notebooks, protocols and reading notes into one shared, searchable wiki.

  • Project documentation

    Bring a project’s scattered specifications, reports and email threads together in a traceable way.

  • Onboarding and handovers

    Turn knowledge locked in inboxes and file shares into something newcomers can query themselves.

  • Archives and collections

    Open up large, heterogeneous holdings – with cross-references that do not exist in the raw material.

Technology

The architecture at a glance

Four layers, deliberately simple: raw data stays untouched, and everything generated is readable and versionable Markdown.

01

Compiler

Python

Reads raw data, calls the language model and produces wiki pages, entities and links.

02

Storage

Markdown · JSON · Graph

Pages as Markdown files, knowledge graph and state as JSON – no separate database.

03

API

Node.js · Express · TypeScript

Authentication, pipeline control, retrieval and chat endpoints behind a lean REST interface.

04

Interface

EJS · Tailwind CSS

Server-rendered, bilingual (DE/EN) and built for small servers and small screens.

Reproducibility

Traceable and lean to run

The whole prototype runs in a few containers with fixed resource limits and can be reproduced with Docker Compose – even on a small server. Evaluation scripts for retrieval, faithfulness, entity resolution and PII redaction are part of the project.

  • Self-hosted, with no trackers or external fonts
  • Reproducible setup with Docker Compose
  • OpenAI-compatible or local language models
  • Built-in evaluation harnesses

FAQ

Frequently asked questions

Is Wissensbau a product?

No. It is a non-commercial research project. There are no prices or sales plans, and no study results exist yet.

Where does the idea come from?

From the LLM-wiki pattern described by Andrej Karpathy: a language model maintains a wiki built from raw material instead of searching afresh for every question. We study how well that works in practice.

Can I try the system?

Access is limited to test accounts so the small server stays reliable. Write to us via the contact page if you are interested.

What happens to my documents?

Raw data stays untouched on the server. For processing, text excerpts are sent to the configured language-model provider – or not sent outside at all when a local model is configured. The prototype’s sample data is fictional.

Can I use the code?

Yes, the source code is public on GitHub. The setup is reproducible with Docker Compose, even on small hardware.

How are the answers checked?

Through evaluation scripts for retrieval, faithfulness, entity resolution and PII redaction. We only publish results once an evaluation supports them.

About the project

A project by Navid Falah

Wissensbau is being built as a research and learning project around LLM-assisted knowledge systems. Feedback, questions about the method and reports of errors are always welcome.

Get in touch

Try the prototype

Access is limited to test accounts. Sign in to explore the pipeline, wiki and chat with sample data.