Case Study

Adaptive reasoning beyond LLM’s


AI is in significant evolution, with emergent capabilities and potential to outstrip human cognition. However current approaches to AI have significant limitations.

The Challenge

Large language models have significant limitations. They are fixed on inputs and outputs and need brute-force training. They have no true understanding of what they are doing, are expensive to run, and only address shallow use-cases (question – answer; or single turn prompt – output).

GEAR (Goal Event Alignment and Resonance) has significant potential to solve some of the challenges in generative AI:

  • Hallucinations can be replaced with true understanding,
  • Brute force learning can be replaced with more efficient use of data and compute,
  • Single turn answers (prompt – output) can be replaced with multi-step problems and complex system management,
  • ‘Same result for everyone’ can be replaced with human empathy, and a deeper understanding of the human and environmental context.

How Akin Helped

We are developing a novel approach to Generalizable Artificial Intelligence. Unlike classic statistical learning like large language models, Akin’s AI has great potential to autonomously solve complex problems in unpredictable environments, and form deep relationships with humans.

As a platform with a frontier model, we have focussed on several key differentiators:

  • Complex systems management and analytics,
  • Problem solving and task performance,
  • Human interaction with greater state-awareness.

We have benchmarked and deployed our AI across a number of sectors, including a Personal AI to support daily living, health management, ecosystem modeling, robotics, Space Industry, advanced manufacturing  (AI and robots working alongside humans in high-compliance environments), complex task support and adaptive reasoning, social companions, ambient AI systems to run a habitat or environment, and complex systems management.

We are benchmarking our AI systems for future roles in human cognitive augmentation, autonomous adaptive reasoning, complex task support, ability to operating with high efficacy in environments with minimal data dependency; and potential for future alternate computing systems like Quantum Computing.


“[Akin AI]... is not limited to single-turn prompts, but can handle complex problems. The AI must work in an environment without cloud compute resources, so it has been designed to be more efficient and less reliant on brute-force learning. This is a significant  development given the costs of running generative AI foundation models.”

Australian Space Agency, January 2023