Everything happens in the e5o dashboard — you bring a research idea and an AI-model API key.
Open the app and sign in with the email and password you were invited with. Access is invite-only — if you need an account, ask your administrator.
In the sidebar, open ⚙️ LLM Settings:
① Pick a provider (Anthropic, OpenAI, OpenRouter, Ollama Cloud, Gemini, or DeepInfra)
② Paste your API key and click “Test key & load models”
③ Pick a model from the live list (use the search box) and click Save
④ Click “Test chat” — you should see the model reply
Tip for your first run: choose a capable but affordable model — a full paper run makes hundreds of model calls.
Open 🧙 Wizard, type your research topic in plain English — for example “Domain generalization under distribution shift” — and step through the prompts to launch the run. That's the last required input: the pipeline is autonomous from here.
The 23-stage pipeline runs on its own: literature review → hypothesis (multi-agent debate) → experiment design → real code execution → analysis → peer review → paper writing.
Follow live progress in 📊 Dashboard and 🔬 Pipeline, or ask questions in 💬 Chat. Long runs are normal — hours, not minutes.
When the run finishes, open 📄 Paper to read the result. Deliverables include compile-ready LaTeX, BibTeX references, the experiment code, and charts — everything you need for a submission.
Sign in → connect a model → give it a topic → wait → collect your paper. Every later run is just steps 3–5.