Introducing the Artificial Intelligence Toolkit that makes AI simple for critical decision-making

Traditional AI solutions, such as deep learning, often lack transparency and explainability, making them unsuitable for aerospace, defense, healthcare, and other high-risk industries. Thales' AI Toolkit overcomes these limitations by providing an interpretable, computationally efficient, and formally verifiable AI framework, making AI safer and more reliable for critical operations.

Combining explainability, efficiency, and trust, Thales' AI Toolkit brings a new level of intelligence to mission-critical decision-making across industries.

The Thales AI Toolkit simplifies AI for mission-critical applications by combining Genetic Fuzzy Tree technology, explainability, and formal verification. It enables trustworthy, low-power AI that supports predictive analytics, autonomous operations, and real-time decision-making.

Genetic Fuzzy Tree: Human-Inspired, Transparent and Verifiable AI

At the core of Thales’s AI Toolkit lies a unique machine learning process known as the Genetic Fuzzy Tree. This algorithm breaks down complex problems hierarchically and in a non-binary, human-like manner by making sub-decisions whenever possible. Originally developed by the US startup Psibernetix, which joined the Thales Group in 2018, the technology showcased its capabilities during a 2016 US military project. In this high-fidelity simulation, the system successfully controlled unmanned combat aerial vehicles in air-to-air combat missions, outperforming several renowned fighter pilots and tacticians. When results were published, the story quickly went viral on social media.

Explainable, Certifiable AI for Safety-Critical Systems

According to Nick Ernest, Thales expert and originator of the Genetic Fuzzy Tree methodology, “our aim is to take something intrinsically complex and make it simple.” Unlike conventional deep learning approaches, this AI is entirely transparent, operates in the linguistic domain, and is based on mathematically accessible foundations. This allows every decision to be logically explained and the entire process to be verified, validated, or certified — a major advantage for safety-critical environments. While matching the performance of other AI methods, Thales delivers key benefits: low-power consumption, high efficiency, and trustable AI for safety-driven markets.

AI Toolkit: Three Tools, Three Flexible Service Levels

Thales has fully integrated the Genetic Fuzzy Tree methodology into its innovation roadmap, notably through COMBI, a bidirectional operator/system translator awarded the European Defence Agency’s Defence Innovation Prize in 2021. COMBI is the first concept to embed human intentions within a system, translating high-level operator goals into dynamic interaction with multiple intelligent subsystems throughout a flight mission. This enables pilots to assess real-time information and make the best decision with full confidence.

This philosophy is embedded in the Thales AI Toolkit, structured around three complementary components:

1. A software package for building fuzzy logic-based smart agents, processing data from inputs to platform control.
2. An intelligent optimization environment to enhance agent performance.
3. A formal verification tool to ensure full compliance across all data flows and processes.

To meet a wide range of customer needs, Thales also offers three levels of service:

• Level 1: the customer builds their own AI application using the Toolkit under license, with Thales providing training and support.
• Level 2: co-development between Thales and the customer.
• Level 3: full application development managed entirely by Thales.