# AI Research Brief Pipeline

A compact Python pipeline for turning a small set of research notes, articles, source excerpts, or interview snippets into a structured executive briefing.

This is a public-safe demonstration project. It uses fictional demo material and does not require external AI APIs, which makes the workflow reproducible and easy to inspect.

## What It Does

- Loads research sources from JSON
- Extracts recurring keywords and themes
- Selects high-signal excerpts
- Groups findings by strategic relevance
- Generates a Markdown executive briefing
- Provides a CLI for repeatable research workflows

## Why This Exists

Strategic advisory work often starts with fragmented source material: articles, meeting notes, market observations, public reports, and expert comments. This project demonstrates how that material can be converted into a clear briefing workflow before deeper human analysis or optional AI augmentation.

The goal is not to replace judgment. The goal is to create a reliable first-pass structure that reduces cognitive load and improves decision preparation.

## Quick Start

```bash
python3 -m venv .venv
source .venv/bin/activate
python -m pip install -e ".[dev]"
research-brief examples/demo_sources.json --output examples/demo_brief.md
```

A generated reference report is included at `examples/sample_brief.md`.

## Project Structure

```text
src/ai_research_brief_pipeline/
  __main__.py      CLI entry point
  models.py        Typed source and briefing models
  analysis.py      Keyword, theme, and excerpt analysis
  reporting.py     Markdown briefing generation
examples/
  demo_sources.json
  sample_brief.md
tests/
  test_analysis.py
```

## Use Cases

- Executive research briefings
- Innovation scouting notes
- Market observation summaries
- Technical due diligence preparation
- Podcast guest research
- Thought leadership source synthesis

## Development

```bash
source .venv/bin/activate
python -m pip install -e ".[dev]"
python -m pytest -q
```

## Roadmap

- PDF export
- CSV source ingestion
- Optional OpenAI/Claude summary layer
- Source weighting by credibility
- Web article ingestion
- Notion or Google Docs export

## License

MIT