A product studio, not a monolithic platform.
AnimusLab builds focused, independent infrastructure engines. Every product is standalone, independently deployable, and solves a first-principles problem. They interoperate through clear contracts and dogfooding — not forced platform dependencies.
- Products
- 5
- Model
- Independent engines
- Coupling
- Optional contracts
Select an engine to inspect.
Each system carries its own repository, release cadence, and scope boundary. The matrix below records where one tool consumes another — and confirms none of those relationships is required.
Anchor Engine
Deterministic static analysis and sub-millisecond runtime statutory enforcement.
When developing the federated training workflows in AnchorGrid-Hub, participating institutions could fine-tune base models on proprietary data and submit adapter weight merge requests back to the network. This raised two critical questions: How do we mathematically verify that incoming weights improve capabilities rather than degrading performance, injecting backdoor triggers, or causing alignment drift? And how do we deterministically govern runtime AI tool execution and model outputs against statutory requirements (EU AI Act, RBI FREE-AI, SEC Reg SCI) with zero-latency overhead? Anchor was built to answer both.
- [1]
Zero-copy Rust static analysis AST engine parsing code and prompt definitions against formal .anchor dialect rules.
- [2]
Sub-millisecond (< 0.4ms) synchronous intercept runtime gate enforcing hard statutory constraints before LLM calls execute.
- [3]
Decision Audit Chain (DAC) generating SHA-256 Merkle-linked, Ed25519-signed immutable compliance records.
- [4]
Open-core architecture: local engine is 100% free and open-source; anchor-web provides enterprise SaaS governance.
- ✓An open-source CLI and Rust/Python SDK for AI governance.
- ✓A static scanner that blocks compliance violations in CI/CD pipelines.
- ✓A runtime guard enforcing legal and safety boundaries before execution.
- ✕A post-hoc monitoring dashboard (that is anchor-web).
- ✕A model training framework.
- ✕A cloud service (the engine runs completely local and air-gapped).
Used as the formal verification gate in AnchorGrid-Hub weight merges, and as the audit engine for QuantForge-AI execution boundaries.
The dogfooding & independence matrix.
Every integration across the studio is an optional consumer contract. Systems run standalone and do not require monolithic omnibus runtimes.
| Host system | Integrated tool | Nature of integration | Operational without tool? |
|---|---|---|---|
| AnchorGrid-Hub | Anchor Engine | Validates incoming LoRA weight merge requests against statutory rules and performance bounds before merging. | Yes — Hub can merge weights using standard loss checks without Anchor. |
| AnchorGrid-Hub | Shadow_Watch | Protects API endpoints and model submission routes with continuous behavioral trust scoring. | Yes — Hub functions with standard API key / bearer token auth. |
| QuantForge Terminal | QuantForge-AI | Embeds quantitative reasoning and natural language financial analysis into the TUI. | Yes — Terminal operates as a standalone scriptable financial workstation. |
| QuantForge-AI | QuantForge Terminal | Uses the terminal's financial connectors (SEC, FRED, MarketWatch) as live execution tools. | Yes — AI agent can connect to external Python or HTTP data feeds. |
| anchor-web | Shadow_Watch | Provides zero-friction continuous session security across compliance and auditor portals. | Yes — anchor-web operates with standard TOTP clearance tokens. |
| AnimusLab Tooling | FORGE | Provides rapid zero-cost persistent state storage during active prototyping sprints. | Yes — Systems can connect directly to PostgreSQL, SQLite, or Redis. |
Guidance for institutional & regulatory evaluators.
If you are evaluating AnimusLab technologies for regulatory compliance or institutional deployment:
1. Governance audit
Inspect Anchor Core & anchor-web for statutory AST compilation, zero-copy Rust runtime hooks, and Decision Audit Chain cryptographic proofs.
2. Security audit
Inspect Shadow_Watch for continuous ensemble scoring, KL-divergence behavioral modeling, and Jaccard entity continuity.
3. Storage audit
Inspect FORGE for .forge binary format parsing safety, in-memory query isolation, and Write-Ahead Logging integrity.
Official specifications & test suites are available in each respective repository.