Skip to main content

​
Why?

Colorful gradient image with the titel /a big clusterfuck

Is it not?

The modern information landscape is a chaotic mess and, most often, difficult to navigate. Maybe you need to prove a point, are aiming to dive deeper into a topic, or simply want to understand what’s happening. Whether you’re an organisation drowning in unstructured data, an NGO with not enough hands on deck, an independent journalist tracking patterns across hundreds of sources, a researcher needing systematic analysis tools, or a business struggling with process inefficiencies, this is for you.”

​
The Idea

A journalist knows how to identify “security framing” in news coverage. A policy analyst knows what counts as “meaningful stakeholder engagement” in legislative proposals. A bureaucrat knows whether a grant application is properly filled out. That expertise lives in their heads, maybe in spreadsheets and notes. This works great for tens of documents. At hundreds or thousands, you’re either stuck or you need to hire engineers. Meanwhile, sophisticated analysis infrastructure, the kind that lets you systematically apply analytical frameworks at scale, has only been available to well-funded institutions. This is where HQ shines: define your analytical questions in plain language. Apply them at scale. Your schemas are your “lenses” you use to see the data. They are shareable, transparent, and improvable. Other researchers can see exactly how you defined your framework, critique it, refine it, or apply it to their own data. For example: Imagine as a journalist, you are analysing 200 news articles. You create a schema:

Schema configuration interface

Schema preview interface
Run it. Get structured data showing systematic patterns. Export visualizations.

Annotation runs

Annotation results

Multi dashboard
The principle scales. A journalist knows “framing.” A policy analyst knows “stakeholder engagement.” An NGO worker knows what signals a policy shift. Schemas let you take that knowledge and apply it systematically across thousands of documents. The same infrastructure that analyses legislation can sort emails, process intake forms, track regulatory changes, or monitor media coverage. This capability shouldn’t be locked behind institutional walls. We’re building it as public infrastructure: schemas, geocoding, vector search, local AI. Basic components, simple when you list them out. But that’s the point. These are basic intelligence capabilities an open society needs, like libraries or archives, and they should be equally accessible. Open source. Self hostable. Bring your own LLM keys if you want privacy. Share your analytical frameworks publicly if you want transparency. Use it for journalism, research, advocacy, governance, anything that serves the public interest. Importantly: none of this would exist without the many people dedicated to open source. We’re assembling the hard work of countless other projects. We are standing on the shoulders of a massive collaborative ecosystem that makes this possible. We are grateful for their work and we are proud to be part of it.

​
How It Works

1

Ingest Content

Upload files, URLs, search results, RSS feeds
2

Define Schemas

Describe what information to extract in plain language (see above)
3

Run Analysis

Use AI to apply your schema at scale across documents
4

Explore Results

View through tables, visualizations, maps, or export the data

Supported Formats

PDFs, web articles, text, CSV, RSS feeds

Coming Soon

Images, audio, email inbox ingestion
The schemas are the key innovation. They let you formalize your analytical method in natural language, making qualitative approaches reproducible and transparent. Other researchers can see exactly how you defined “populist rhetoric” or “security framing” and apply the same lens to their data. Example annotation schema

​
Chat & MCP

The Chat & MCP (Model Context Protocol) centralizes all platform tools (asset management, schema based analysis, vector search, content ingestion) into a unified analysis access point that lets you work with structured analytical methods through conversational AI. This combination allows for various workflows for researching & synthesizing data. Editing MCP configs (upcoming on a per user level) furthermore allows you to grant your LLM access to almost arbitrary outside data and use it for your analysis.

​
Infospaces & Vector Storage

Your assets, schemas and analysis results are scoped to “Information Spaces” that you can use to curate information. Each Information Space is a dedicated vector space. Use vector embeddings from local or cloud models to search through your data, cluster it (upcoming) and find duplicates.

​
Getting Started

​
Option 1: Use the Hosted Instance

The easiest way to start. We host the infrastructure, you bring your own LLM API keys.
1

Register

Sign up at open-politics.org/accounts/register
2

Add API Keys

Configure your LLM API keys on the home page
3

Start Analysing

Upload content and create your first schema
Your account also works on the forum for community support.

​
Option 2: Self-Host with Docker

For privacy, customization, or institutional requirements. Run everything on your own infrastructure. 
git clone https://github.com/open-politics/open-politics-hq.git
cd open-politics-hq
bash prepare.sh
cp .env.example .env
# Edit .env with your configuration
docker compose up
Default admin credentials (change these): 
FIRST_SUPERUSER=app_user
FIRST_SUPERUSER_PASSWORD=app_user_password

​
Deployment Flexibility

Fully Local

Run everything on your own hardware. Good for air gapped environments or complete data control.

Hybrid

Run the application locally but use managed services (AWS RDS, Upstash Redis, S3) to reduce operational burden.

​
Architecture

The platform is built from several independent services that work together. You can run them all locally or mix local and managed services.

​
Core Components

ComponentWhat It DoesTechnology
BackendAPI, analysis jobs, MCP serverFastAPI + Python
FrontendWeb interfaceNext.js + React
WorkerBackground processing for large jobsCelery
DatabaseData storage with vector searchPostgreSQL + PGVector
Object StorageFile storage for uploadsMinIO (S3-compatible)
Cache/QueueSession management, job queuesRedis
GeocodingLocation extraction and mappingPelias
LLM (optional)Local AI inferenceOllama

​
LLM Support

Connect any of these AI providers:

Anthropic

Claude Sonnet 4.5, etc. (Best Experience with tools)

OpenAI

GPT-5 etc. (Not sure)

Google

Gemini models (Large context, speed and modality capabilities)

Ollama

Run models locally: Llama, OAI OSS, Qwen, etc. (Best Privacy)
Configure API keys in the web interface or run Ollama locally for complete privacy.

​
Contributing

We’re building this in the open. The codebase, analytical methods, and documentation are all public and improvable.

Code Contributions

Report bugs, suggest features, contribute code

Documentation

Improve docs, add examples, share schemas

Community

Join forum discussions, share knowledge

Research

Build analytical frameworks, publish findings

​
Contact & Community

Dev Meetings

Wednesdays 15:30 Berlin Time

​
License

AGPLv3, see LICENSE This means you can use, modify, and distribute this software, but any modifications or services built on it must also be open source. You can get an enterprise license for private use modifications which are not publicly deployed for one year at a time under strict ethical guidelines.

​
Project Origins & Story

Talk: Open Source Political Intelligence

CCCB Datengarten Presentation
 This project started at Freie Universität Berlin’s Political Science department. The gap between what research infrastructure could theoretically do and what was practically accessible to most researchers was obvious. Someone had to build the bridge. Our first official funding came from the European Horizon NGI Search project, and before that we were lucky to have a warm circle of friends and early supporters who believed in the idea before there was much to show. No venture capital, just journalists and neighbours who got it. The business model is intentionally simple: we’ll try to keep a free hosted version running as much as we can shoulder, but hosting HQ is rather cheap as most of the cost comes from LLMs. We focus on helping organisations deploy and use the platform effectively: infrastructure orchestration, custom implementations, training. The platform itself always stays open source and self hostable. Our marketing is the research that gets done with these tools. If people publish interesting work using HQ, that’s worth more than any advertising campaign. If you want to support this mission or just talk about what you’re building, reach out: engage@open-politics.org
⌘I