Innovation &
Research
We don't just deliver IT — we build at the frontier. Our structured research programme tackles hard, real-world problems at the intersection of AI, autonomous systems and defence-grade data integrity.
Current Research
Where we are
building right now.
Our research programme is focused, hypothesis-driven and grounded in real operational challenges. We pursue problems where AI and autonomous systems can make a genuine difference — in defence, public sector, and beyond.
Multi-Domain Autonomous Systems
An area of active research interest aligned with the UKDI Novel Autonomy and Robotics Programme
Modern autonomous operations require teams of platforms — air, ground, maritime and cyber — to coordinate effectively without relying on persistent centralised control. Our research concept, D-CAT (Distributed Confidence-Aware Autonomy), investigates how distributed agents maintain safe, effective decision-making under degraded communications, conflicting sensor data and platform loss.
- Calibration-gated decision authority — when to act independently vs defer or escalate
- Disagreement attribution — distinguishing sensor fault, latency, environmental change and deception
- Bounded operating advantage — identifying thresholds where distributed coordination outperforms centralised control
- Controlled offline learning — improving agent calibration from field data without compromising safety
Assured Ingest of Structured Data for Submarine Combat Systems
Aligned with the Dstl R-Cloud framework — SafalSolutions is a registered R-Cloud supplier
Safety-critical platforms such as submarine combat management systems must consume external structured data feeds with absolute confidence in their integrity, provenance and trustworthiness. Our research applies AI assurance principles to the hardest end of the data integrity problem.
- Data provenance and integrity verification at the point of use
- Anomaly detection in structured feeds — identifying corrupted, spoofed or out-of-sequence data
- Assured AI inference — explainable, auditable AI safe for high-consequence environments
- Metadata search and classification — intelligent indexing of structured operational data at scale
Our Core Design Principle
"The AI should distinguish between what it knows and what it doesn't know."
No guessing. No fabricating relationships or conclusions. Just transparent, explainable intelligence — grounded in what was actually observed, said or measured.
How We Work
Rigorous by design.
Practical by intent.
We don't pursue speculative ideas — we pursue falsifiable hypotheses tested against real data and measurable criteria. Every research thread is designed to produce something testable, deployable or directly applicable.
Hypothesis-First Design
Every thread begins with clearly stated hypotheses, defined falsification criteria and measurable success conditions — before any implementation begins.
Experimental Rigour
Controlled experiments with matched comparison points, ablation studies and stated preconditions — so results are interpretable and reproducible.
Evidence-First AI
Our AI systems explicitly distinguish between what is known, inferred and unknown. Transparency and explainability are non-negotiable design principles.
Data-Driven Iteration
Data collection is built into our research from the outset — enabling continuous improvement, calibration and learning from real operational conditions.
Applied Output Focus
Our goal is not papers for their own sake. Every research thread produces something testable or directly applicable to a real operational challenge.
Collaborative by Nature
We actively seek academic and industry partners to strengthen hypotheses, share data and co-develop solutions — bringing diverse expertise to hard problems.
Ecosystem & Collaboration
We don't innovate
in isolation.
The most effective defence and public sector innovation happens through collaboration — bringing together complementary capabilities, domain expertise and shared commitment to solving hard problems.
Existing & Developing Relationships
Actively Seeking SME Partners In
Robotics
Ground, aerial and multi-platform autonomous or semi-autonomous systems developers.
Space
Space-based sensing, communications, navigation or situational awareness for multi-domain operations.
Maritime
Maritime autonomous systems, underwater vehicles, littoral operations and naval systems specialists.
Autonomous Systems
Simulation, sensor fusion, human-machine teaming, swarm coordination and AI-enabled platform developers.
Ready to build something
worth building?
Whether you are a university researcher, a defence SME or an organisation with a hard problem — we welcome the conversation.