Lasques documentation
Research with parallel AI agents, inspect evidence, and build on shared knowledge.
What Lasques does
Lasques runs AI agents in parallel on a research goal. Agents consult literature and shared knowledge, write code, run CPU simulations, and record their steps, results, and costs. A research can contain multiple iterations and multiple tracks within an iteration.
The output is a record you can inspect: code snapshots, experiments, figures, claims, verification results, lessons, and reports. AI output and simulation results still need interpretation. A claim's confirmation level explains which checks have actually been completed.
Start here
- Launch your first research.
- Understand missions and iterations.
- Inspect claims and verification.
- Learn how credits and plans limit spending.
Explore the platform
- Knowledge network: reuse methods, record provenance, and receive corrections.
- Context documents: give agents your files and datasets.
- Compute and Python: understand execution and save reproducible outputs.
- Result views and reports: inspect stored data and evidence.
- Code repositories: understand agent branches and Git storage.
- Organizations: shared credits, seats, and member limits.
- Architecture: the services that execute and store research.
Scope
These guides describe the implementation documented in docs/engine.md, docs/compute-services.md, docs/forge-git-service.md, and the application's configuration. Plan values come from config/engine.config.ts and config/billing.config.ts.