Composabl
Revolutionize industrial automation with intelligent, no-code AI agents.. [Contact for Pricing]
Last verified:
What is Composabl?
Composabl is a multi-agent AI production platform that empowers engineers to create AI agents and agentic systems for complex industrial processes. Built by ex-Microsoft engineers and led by industry trailblazer Kence Anderson, the platform enables engineers to directly teach AI agents to work alongside them in real-world settings by leveraging operator expertise to guide the development of autonomous automation effortlessly.
The platform features a unique two-part architecture combining a no-code UI with a Python SDK and CLI, giving users both usability and power. Composabl agents make decisions based on perception, much like humans would, and can manage diverse equipment from CNC machines and bulldozers to drones, robotic arms, and chillers. The platform uses Machine Teaching methodology that utilizes the knowledge of process engineers and industrial engineers to revolutionize industrial automation.
Key capabilities include simulation-based training with deep reinforcement learning, real-time physical deployment connecting to SCADA, MES, HMI, and ERP systems, an Orchestration Studio for building multi-agent systems, and the ability to import ML models, LLMs, and custom algorithms as modular building blocks. Agents are trained in realistic simulations before deployment and can be benchmarked and compared.
Composabl is designed for industrial engineers, process engineers, subject matter experts, and technologists working in manufacturing, transportation, healthcare, and facility support systems. The platform helps optimize production yields, enhance asset routing, facilitate patient flow management, and reduce energy usage through AI-driven automation that delivers real-time intelligence for industrial operations.
Composabl pricing
Pricing model: Freemium
Composabl annual licenses and pilots are available through Microsoft Azure subscription. Three tiers based on user count: Starter Tier (5-9 users) includes 1 platform seat, 1 SDK access license, 1 agent deployment per user, and 10,000 free training credits per user. Builder Tier (10-24 users) includes 1 agent deployment per user and 15,000 free training credits per user. Pro Tier (25-99 users) includes 1 agent deployment per user and 20,000 free training credits per user. Each user receives login access to the Composabl No-code UI and SDK. Custom offerings available for team trainings, custom consulting, and other platform needs. Two-day Machine Teaching methodology training available.
Composabl pros
- No-code UI enables engineers without programming backgrounds to create agents
- Python SDK provides advanced flexibility for AI R&D and enterprise deployment
- Machine Teaching methodology leverages operator expertise directly
- Agents make perception-based decisions like humans do
- Deep reinforcement learning training in realistic simulations
- Real-time physical deployment to control real equipment
- Integrates with SCADA, MES, HMI, and ERP systems
- Modular architecture with interchangeable, upgradable parts
- Explainable AI with traceable decisions to agent behavior models
- Multi-agent systems that scale industrial operations
- Import custom ML models, LLMs, and algorithms as building blocks
- One-click agent training at scale from the UI
- Compare and benchmark multiple agent performance before deployment
- Available through Microsoft Azure subscription
- Collaborative team features for working with co-workers
- Preserves and scales human expertise through intuitive interfaces
- Adaptable agents that evolve based on changing operational conditions
Composabl cons
- Requires minimum 5 users for Starter Tier pricing
- Annual licenses only, no monthly subscription option
- No free tier available for individual testing
- Requires Python 3.10.x for SDK and CLI
- Limited to 1 agent deployment per user per tier
- Training credits capped based on tier (8,000-20,000)
- Complex platform requiring expertise to deploy effectively
- Requires simulation setup before real-world deployment
- Pricing only available through Microsoft Azure subscription
- Custom offerings required for team training and consulting
Frequently asked questions about Composabl
What is Composabl?
Composabl is a multi-agent AI production platform that empowers engineers to create AI agents and agentic systems for complex industrial processes. It is a workbench for building industrial-strength intelligent agents capable of making high-value industrial decisions. The platform enables engineers to directly teach AI agents to work alongside them in real-world settings using Machine Teaching methodology.
How does Composabl differ from traditional AI chatbots?
Unlike chatbots that use Large Language Models to create paragraphs or pictures, Composabl uses LLM to create skills and goals for decision-making agents that can control equipment and processes in real-time. Composabl agents make decisions based on perception and feedback, managing complex operations like robotics, drones, and industrial machinery in the physical world.
What equipment can Composabl agents control?
Composabl agents can manage a diverse range of equipment including CNC machines, bulldozers, drones, robotic arms, chillers, manufacturing equipment, and various industrial machinery. They connect to control systems and interfaces like SCADA, MES, HMI, and ERP for real-time physical deployment.
What is the Machine Teaching methodology?
Machine Teaching is Composabl's approach that utilizes the knowledge of process engineers and industrial engineers to revolutionize industrial automation. Expert operators can teach AI by embedding their experience into agent behavior through the no-code UI, where they define goals for skills and the platform turns these into reward functions for deep reinforcement learning training.
How do I get started with Composabl?
First, create a new project containing all agents for one use case. Select the simulator you will use and select your team for collaboration. Tell Composabl what you want your agent to do and the platform creates the starting point. Choose how to train it and watch it improve. You can train multiple agents, compare performance, and export the trained agent to connect to real equipment for deployment.
What is the difference between the no-code UI and Python SDK?
Composabl is a two-part platform with a no-code UI and Python SDK/CLI. The no-code UI enables subject matter experts and process engineers to create teams of agents from modular building blocks and train them without programming. The Python SDK allows advanced users to create agent components like skills, ML models, LLMs, and algorithms, then publish them to the UI with one CLI command.
How does deep reinforcement learning training work in Composabl?
Agents learn by practicing in simulation and getting better through deep reinforcement learning. When creating a skill in the UI, you select goals and Composabl turns these into reward functions. Goals apply to sensor variables using directives like Avoid, Maximize, Minimize, Approach, or Maintain. The agent trains through experimentation until ready to control real systems.
Can I import my own machine learning models?
Yes, engineers can import machine learning models, LLMs, controllers, and optimization algorithms as building blocks. Using the Python SDK, you can create or package ML models to add advanced perception, LLMs to add natural language, and controllers for programmed decision-making, then publish them to the UI with one CLI command.
How do I deploy agents to real equipment?
Once your agent is trained and you've decided which one to deploy, export the trained multi-agent system and connect it to the Composabl runtime for deployment. The agents then connect to your real equipment, control systems, and interfaces like SCADA, MES, HMI, and ERP to perform in real-world scenarios.
What industries benefit from Composabl?
Composabl optimizes performance across manufacturing (boosting production yields), transportation (enhancing asset routing), healthcare (facilitating patient flow management), and facility support systems (reducing energy usage and expenses). The platform serves any industry with complex industrial processes requiring real-time AI-driven decision-making and automation.