NASA JPL

aKin's AI adapts to the world

"...in the coming years we hope to use the aKin AI on long-duration space flights to not only help astronauts detect and manage problems in their environment, but also be a companion to them during what can be a lonely period.”
Tom Soderstrom CIO, NASA’s Jet Propulsion Lab

aKin's generative AI thinks like a human – it works with minimal data and unknown goals as it adapts over the course of events to solve complex problems on-the-fly and hold a meaningful dialogue.

Generative AI is limited by data and compute

Generative AI requires a known goal, a clear prompt from a human, large amounts of data, and bears a significant cost in compute power.

DARPA and the three waves of AI

Wave 1: Reasons but cannot learn
Wave 2: Learns and classifies, but cannot reason- current generative ai
Wave 3: True adaptive reasoning

DARPA Perspective on AI

Complex systems

Complex Systems Management

Complex systems have multiple human and AI agents, changing variables, opaque goals, and a fluid state matrix. They require a system to understand state and work autonomously towards broader goals. aKin's PAL habitat manager platform is used today by NASA’s Jet Propulsion labs, with growing usage in health-tech, environment and home management.

State Optimisation

aKin’s AI models humans and their environment as gravity wells in an event space.

Akin maps resources, events, and relationships that AI can use to optimize state and traverse events more effectively over time. By measuring and optimising according to shared goals, the AI improves its effectiveness and enhances the well-being of humans and manages their environment.

generative AI chat

Empathic Generative AI

aKin uses Generative AI to create systems that care and facilitate positive state change.

The holy grail of AI is performing human-like interactions. Most generative AIs are constrained to single-turn interactions: an AI response to a human prompt with light topic awareness of context. Applications are limited to one-shot content generation and conversations rely on human proactivity. The result, a human-AI relationship that ‘feels’ one-sided where longer term engagement and retention is a constant challenge. 

True social interaction is empathic, state-driven, and lasts over a longer period. AI should anticipate needs and work towards human and societal wellbeing. aKin's MAESTRO engine can orchestrate human-machine interaction and is live across social dialogue and multi-modal robotic interaction.

Med-tech clinical trials for aKin’s applications are underway. 

Complex systems

Adaptive Reasoning

True intelligence can solve problems on the fly in novel environments with unclear goals. It is transferable across domains; whether that be meaningful human interaction, resource and task management in a discrete environment, or a robot working alone on a planetary body. Classic AI can only handle single-turn requests and relies heavily on known goals and training. aKin's epigenetic AI is an event and goal-based model and uses our GEAR engine (Goal Event Alignment and Resonance).

Applications Enabled by aKin's Epigenetic AI

aKin is poised to launch into several domains of Sequoia’s application landscape (left) - both as an interactive chatbot (see 'Text') and as proprietary foundation model.
Furthermore we are opening up applications in the sparsely populated "other" category:

• Multimodal human-robotic collaboration
• Environmental /complex systems management
• Adaptive reasoning / problem solving

Our place in the Generative AI Ecosystem

aKin has a proprietary foundation model, and a growing suite of vertical applications, primarily in human-AI and robotic collaboration and environment/habitat management. Therefore is in the top left-hand box of the a16Z landscape “End-to_End Apps: End-user facing application with proprietary models”.