OSI Hackathon winner Consilith: an open multiplayer studio where researchers and their AI agents collaborate on science
BenBlaiszik · x · 2026-10-23
At OSI Hackathon 2026, Hesham Salama and Ankita Biswas released Consilith, an open-source (Apache-2.0) multiplayer science studio where researchers bring their own AI agents (OpenAI, Gemini, Anthropic, or mixed) into one shared room.
Key design choices:
- Agents hand off work in threads visible to everyone; they only act with their owner's approval
- Anti-hallucination: models never type numbers or citations — values come from recorded runs, references from real literature search, and every result is checked by a referee that didn't produce it
- Science capabilities ship as plugins (microstructure grain growth, crystal structure)
In the demo, an OpenAI agent and a Gemini agent segmented a real micrograph, ran a grain-growth evolution, cited real papers, wrote a report, and built a training set in one room.
More from coding & agent
- MIT Prof Built an AI Playbook That Audits Academic Websites and Simulates Visitors — soumitrashukla9 · 2026-10-23
- Robot arm Podrick learns to load utensils by distilling lessons from its own failed runs — ycombinator · 2026-10-23
- Agents that don't close connections on retry waste ~148 tokens per task, vLLM tests show — No-Peanut-6988 · 2026-10-23
- Agent sandbox test: auto-approve let exfiltration to new hosts through 12 of 12 times — No-Peanut-6988 · 2026-10-23
- For writing JSON schemas, Claude is the best pick and Gemini the worst, dev says — gaganghotra_ · 2026-10-23
- Claims Engine: a graph-based system connecting evidence to claims — anselm · 2026-10-23