Subconscious Loop

OpenHuman's subconscious is what separates it from every other AI agent. Most AI systems follow the conscious mind model: they wait for instructions and execute them. They cannot think on their own. OpenHuman can.

The Purkinje cell inspiration

In the human brain, there is a specialized neuron called the Purkinje cell. It is located in the cerebellum and is mainly responsible for random thought generation. It plays a significant role in human consciousness.

The human brain operates with both a conscious and subconscious mind. The conscious mind handles deliberate, focused tasks. The subconscious processes information in the background, making connections, surfacing memories, and generating the "random" thoughts that often turn out to be the most important ones. Most agentic AI systems only model the conscious mind. They are reactive. They wait for input.

OpenHuman models both.

How the subconscious works

Inspired by the Purkinje cell, OpenHuman uses Neocortex to periodically trigger global memory recalls. Rather than targeted retrievals in response to a specific query, these are broad, semi-random sweeps across the knowledge graph, pulling up memories based on a mix of recency, relevance, and randomness.

These recalls feed into a self-learning loop. The subconscious processes the recalled memories, looks for patterns, contradictions, and connections, and produces one of several outputs: a proactive insight, an action recommendation, a confirmation of an existing pattern, or a new connection between previously unrelated information.

This loop runs over 10,000 times per day for less than $1. Memory recalls are cheap because Neocortex operates without LLM dependency. The subconscious can think constantly without blowing a hole in your budget.

What the subconscious produces

The subconscious does not generate generic AI outputs. It produces context-specific intelligence grounded in your actual data.

It might surface a risk you had not noticed: two team members making contradictory commitments in separate conversations. It might connect a trend in your trading data to a discussion from three weeks ago. It might remind you of a commitment you made that is approaching its deadline, before you think to ask.

The outputs are proactive. You do not need to prompt OpenHuman to get value from the subconscious. It works in the background, feeding insights into the system that are available when you interact.

The mirror test

The team uses an internal success metric called the "mirror test." When you read the subconscious output, it should feel like looking at yourself in a mirror. Not generic AI analysis. A reflection of your own thinking patterns, priorities, and concerns.

If the subconscious surfaces something that feels alien or irrelevant, it has failed. If it surfaces something that makes you think "I was just about to look into that," it has succeeded.

What the subconscious is not

The subconscious is not AGI. It does not have general intelligence, self-awareness, or autonomous goals. It is a meaningful architectural step toward more human-like AI, built on solid neuroscience principles, but it operates within defined boundaries.

It does not take irreversible actions without your approval. It does not override your explicit instructions. It does not learn objectives that you have not implicitly demonstrated through your data patterns.

The subconscious is a tool for augmenting your thinking, not replacing it.