Agents of
Intelligence

We specialize in optimizing agentic coding pipelines, finetuning powerful AI models, and researching the cognitive intersection between artificial intelligence and dyslexia.

Test Model → Get in Touch

XeLSeD.aI

intelligence amplified

Core Competencies

Pushing the boundaries of automated software development through rigorous research and optimization.

⚡

AI Coding Finetuning

We create specialized datasets and training pipelines to enhance LLM performance on complex software engineering tasks, reducing hallucination and increasing syntax accuracy.

🔄

Agentic Efficiency

Our research focuses on exploiting efficiencies in agentic loops - minimizing token usage while maximizing reasoning depth for autonomous coding agents.

🧬

Pipeline Exploration

We prototype novel architectures that combine retrieval-augmented generation (RAG) with deterministic code analysis tools.

AI & Dyslexia

Bridging Cognitive Differences with Artificial Intelligence

A primary pillar of our mission is researching the interaction between AI agents and users with dyslexia. We believe AI can act as a powerful cognitive equalizer in software development.

✓ Analyzing how syntax highlighting and code structure affect dyslexic readability in AI outputs.
✓ Developing "Dyslexia-Adaptive" system prompts that adjust complexity and formatting dynamically.
✓ User testing interactions to reduce cognitive load during debugging sessions.

Educational Research Projects

Experimental applications exploring early AI model integration, automated content pipelines, and human-AI interaction patterns.

🦎

Invasive Animals

A VR educational experience demonstrating real-time LLM integration pipelines. Features voice-controlled AI gameplay where users teach the model about invasive species through interactive Q&A loops - a testbed for studying agentic learning patterns and human-AI knowledge transfer.

invasivegame.com →

🐾

Animals 1975

An experimental project investigating automated content generation pipelines and early-stage AI model architectures. Explores how foundational AI techniques from the 1970s inform modern finetuning approaches - bridging historical machine learning research with contemporary LLM development.

animals1975.com →

These projects serve as live research environments for testing AI integration patterns, automated pipelines, and model behavior - directly informing our finetuning frameworks and agentic system development.