Choose a generated manuscript from the notebook list to open a submission-style reading page.
DEEPGRAPH · AI RESEARCH HARNESS
An AI research harness that compounds experience
DeepGraph links experiments, knowledge, and evidence into one verifiable loop that keeps improving itself.
LATEST CONCLUSION
—
Research map and full counts
Ordered by paper count. Verdict coverage is the share with and without a conclusion; open questions are sub-problems with papers but no conclusion yet.
Coordinates are pre-computed and inlined, so opening this costs no request and no layout work.
FULL COUNTS
ACTIVITY
Research Runtime
A live view from literature and hypotheses to experiments and evidence-backed decisions.
Research Area Explorer
Method x Dataset Matrix
Adjudicated findings
Manuscript Library
Generated Paper Ideas
No deep discoveries yet. Discovery runs in fixed automatic mode.
EXPERIMENT OUTCOMES
How experiment runs ended
Process Timeline
Chronological record for one agenda: signals, candidate decisions, resource authorizations, jobs, evidence transitions, and verdicts. Failures remain visible beside successes.
No process events recorded yet.
Selection Rationale
why work was chosenNo selection records yet.
Automation Services
Read-only automatic modeEnabled or ready means the service is available; working means a job is executing now.
Idea Experiments
No experiment ideas are active yet. The automatic queue will start them when ready.