Today, Bravo 01 Labs & Dynamics Inc. released DIM-1, its first Decision Intelligence Model (DIM) — a new model category designed specifically for structured decision intelligence rather than conventional language-model generation.
A New Model Category
DIM-1 is not a general-purpose conversational language model. It is a specialized Decision Intelligence Model that transforms structured or unstructured inputs into controlled decisions — decisions that can subsequently be evaluated against explicit rules, authority, and execution constraints.
The architecture is built on a core principle: intelligence and execution governance should be distinct layers.
DIM
Decision intelligence — probabilistic judgment over structured and unstructured inputs.
Korda AI
Deterministic action governance — constraining what actions can be expressed or executed.
This separation allows a probabilistic intelligence system to make decisions while restricting what actions can ultimately be expressed or executed — a design intended for high-consequence environments where control matters.
DIM-1 Results
In its current evaluation, DIM-1 achieved:
- 92.37% held-out test accuracy on BANKING77
- Deterministic reproducibility across repeated runs
- 95.8% top-1 coverage across tested security-sensitive cases where authority changed the governance outcome
- Improved validation accuracy across five training epochs
These results are an initial demonstration of the DIM architecture. Bravo 01 Labs is continuing evaluation across additional datasets, decision domains, security-sensitive scenarios, and governance conditions.
Korda AI and Government-Ready AI
The DIM-1 release expands Bravo 01 Labs' broader AI infrastructure strategy. The company's Korda AI is an assessed Awardable solution through the Department of War's Tradewinds Solutions Marketplace — the U.S. government's marketplace for assessed AI/ML, digital, and data solutions, providing a pathway to government procurement.
Korda is designed as a deterministic governance layer for autonomous and AI-driven systems, using formal action constraints and permissions to prevent unauthorized actions from being expressible within the governed action language.
Bravo 01 Labs is now pursuing an Awardable implementation pathway for DIM, extending the company's government-oriented AI architecture beyond action governance into decision intelligence.
The launch of DIM-1 and the creation of a new category of model, Decision Intelligence Models, is an important step in building the next generation of AI in America. DIM-1 is only a small glimpse of what this architecture can become. We see it as a stepping stone toward a much larger intelligence developments we are working on at Bravo 01 Labs — Developmnents designed towards decision intelligence, governed capability, and systems that can operate with greater determinism and control. What we are releasing today is the beginning, not the destination.
Albert BravoFounder & CEO, Bravo 01 Labs
Beyond DIM-1
DIM-1 is the first public release of Bravo 01 Labs' Decision Intelligence architecture. The company is developing a broader AI infrastructure stack combining decision intelligence, deterministic action governance, sovereignty verification, modular intelligence, and locally deployable AI systems.
The longer-term objective: AI systems capable of operating across enterprise, government, research, and other high-consequence environments — where conventional generative AI architectures may not provide sufficient control over decisions and actions.