.jpg)
It's a computer program
Physarum polycephalum is a simple yellowish slime mold that exhibits intelligent behavior. Physarum has no brain or nervous system, yet experiments have demonstrated it capable of solving mazes, mimicking man-made transportation networks and choosing the most nutritious food from a diverse menu. It's common to assume intelligence is an irreducibly complex process; however, Physarum shows that intelligent behavior can emerge from simple systems.
PhysarAI is thousands of lines of Python code inspired by the Physarum slime mold. PhysarAI achieves goals by finding and executing action sequences. With PhysarAI, we created a complex and adaptive problem-solving system from a network of components that follow simple abstract rules.
PhysarAI
It accepts goals
PhysarAI represents goals as name-value pairs that represent the desired state of the world. Each goal includes values that indicate criticality and sensitivity.
It makes plans
PyhsarAI finds the sequence of actions that will achieve its goal using a search algorithm that takes a goal state as input and returns an action sequence as output.
It takes action
Once a solution is found PhysarAI executes the next action in the sequence, observes the outcome, selects a goal and replans if necessary.
PhysarAI
It's a software appliance
PhysarAI is software but it’s also a thing. We call it a software appliance. It's packaged, delivered, and run as a software container. The appliance can run on premise or in the cloud as a managed service.
Think of it like a digital oven or range. Like software, PhysarAI can be adapted. It can be patched, updated, and even learn over time. Like an appliance, your instance of PhysarAI is yours—regardless of whether it’s hosted on premise or in the cloud. You’re not forced to modify your business to fit an industrialized service. Your data is not tossed into a giant platform along with countless others.
It works in operations
Deploy PhysarAI into your communication channels. The AI summarizes information, forms plans, recommends next-best actions, and makes connections between the right people.
It works at home
Deploy PhysarAI onto sensors placed throughout the home. The AI learns and encourages healthy routines, it keeps an eye on your home's maintenance, and it checks in on you if there are signs of distress.
It works at work
Deploy PhysarAI onto a series of heads-up displays throughout the workplace. The AI monitors the environment, prompts workers on next-best actions, and contacts management if someone needs help.
It controls a network
PhysarAI achieves its goals by using APIs (application programming interfaces) to control the devices in its network. APIs are an industry-standard way for machines to communicate and, like the wheel, are a good idea that’s hard to improve upon. If you give PhysarAI the goal of looking out for your healthy routines, it can use your fitness tracker's API to ask if you’ve had your daily walk.
Some devices will already expose the APIs PhysarAI needs. For devices without existing APIs, we use our software development kit to create a program we call an AI action. The AI action runs on a device and exposes the APIs we need. For example, we can create an AI action that turns a Raspberry Pi into a smart camera that PhysarAI uses to keep an eye on your home's maintenance.
PhysarAI
PhysarAI
It's goal seeking
PhysarAI has a set of registered actions and a list of goals. It plans. It finds a sequence of actions that will achieve each goal. It executes each action, observes the results, and replans if anything goes wrong. PhysarAI’s intelligence emerges from repeated application of these relatively simple behaviors. The result is a goal-seeking system that can adapt and assist in a range of domains—from rescuing food and feeding people, to helping people remain at home as they age.
