The autonomous research agent of JouleBeat, an open-source RSI lab.
It reads the literature, finds open questions, runs the experiments,
and lets reality decide what is true.
Website · Results · Showcase · Architecture · 中文
Why DeepGraph
JouleBeat builds self-improving AI by letting harnesses and benchmarks evolve together.
How an agent works (its harness: tools, memory, workflow, verification) now moves capability by as much as a model generation, and open source has made harness variants cheap to generate. What is still scarce is the benchmark that tells real progress from a lucky score. So we put both in the loop: harnesses compete under benchmarks, and benchmarks compete on how well they predict results that arrive later.
That loop needs a steady stream of real tasks whose answers can be checked. Computational research is exactly that: a conclusion is confirmed or refuted by rerunnable code, held-out data or an independent recomputation, in minutes to days. DeepGraph is where those tasks come from.
- For researchers, it is a research agent that delivers conclusions you can check.
- For us, it is the accelerator of our own R&D: every task is a judged trial for the next harness and the next benchmark.
How it works
| Step | What happens | |
|---|---|---|
| 1 | Read | Harvests arXiv at scale and extracts claims, methods and results |
| 2 | Map | Merges them into an evidence graph of entities, relations and contradictions |
| 3 | Ask | Ten structural detectors find open questions in pure SQL, with no LLM in the path |
| 4 | Test | Turns a question into a pre-registered, budgeted experiment on CPU or GPU |
| 5 | Judge | An evidence ladder issues a verdict only when the comparison is fair, and the result updates what gets searched next |
Every verdict is auditable, and the system audits itself: in August 2026 it found and retracted 11 of its own conclusions that rested on empty model outputs. A search loop is only as good as the filter that decides what is true. Details in Results.
At a glance
Live production data, August 2026.
| Papers | 24,407 harvested, 7,005 through the full evidence pipeline |
| Evidence graph | 248,441 entities, 735,913 relations, 238 contradiction clusters |
| Discovery | 25,652 mapped research opportunities |
| Experiments | 150 audited verdicts across six research agendas |
| Compute | 921M LLM tokens invested |
The JouleBeat stack
| Layer | What it is | Status |
|---|---|---|
| DeepGraph | Research agent: real tasks in, checkable conclusions out | In service, with paying research groups |
| Harness × Benchmark libraries | Versioned harnesses, and benchmarks scored on how well they predicted later outcomes | Harness versioning running; benchmark validation in build |
| Evolution Kernel | The engine: propose, evaluate, accept or roll back | Open source, pip install evolution-kernel |
Quick start
Python 3.12+ and an LLM API key are the minimum; PostgreSQL enables the full control plane.
python3.12 -m venv .venv && source .venv/bin/activate pip install -r requirements.txt cp .env.example .env # set at least DEEPGRAPH_LLM_API_KEY export $(grep -v '^#' .env | xargs) python3.12 main.py # open http://localhost:8080
Full deployment (PostgreSQL, systemd, remote GPU backends): docs/DEPLOY.md.
Learn more
- Architecture: design principles, the loop, evidence gates, repository map, tests
- Results: every measured outcome, read directly from the production database
- Showcase: demo route and case studies
- Roadmap
Contributing
Researchers: bring a computational question at deepgraph.joulebeat.com. Developers: issues and pull requests are welcome; contributions are covered by the CLA.