Documentation
Technical documentation for openPAT — covering its products, technologies, research, projects, developer resources, open-source work, updates, and reference material.
Start Here
Begin with the basics and understand how the documentation is organized.
openPAT
Documentation about the organization and the work surrounding openPAT.
Products
Products and product-specific documentation.
Developers
Technical resources for developers working with openPAT technologies.
Open Source
licensing.
Updates
Changelogs, releases, announcements, and roadmap information.
Reference
Security, privacy, legal, terms, and contact information.
Introduction
Welcome to the openPAT documentation.
This is the central reference for publicly documented work across openPAT. It brings together documentation for products, technologies, research, projects, developer resources, open-source work, updates, and reference material.
The documentation is organized by subject rather than by a single product or system. Each section contains its own technical and contextual documentation, allowing individual projects and technologies to be documented independently while remaining part of the larger openPAT documentation system. What is OpenPAT?
Documentation may change as projects evolve. Pages are updated as systems, interfaces, research, and technical implementations develop.
What is openPAT?
An overview of openPAT, its role, and the work developed under it.
openPAT is the short form used for PheslaX Alpha Technologies, a research, development, and deployment company focused on artificial intelligence and software.
openPAT develops its own technologies, systems, and products, including various artificial intelligence technologies for different applications and use cases.
openPAT is the parent company of Lethan AI. Lethan is one of the products and technology platforms developed within the openPAT ecosystem.
The company's work includes artificial intelligence research, software development, technology development, system deployment, and the development of products and services built around these technologies.
This documentation provides a central reference for openPAT's technologies, products, projects, research, development work, and technical resources. Individual projects may have their own documentation and technical direction while remaining part of the wider openPAT ecosystem.
What We're Building
Explore the products, AI technologies, systems, and research being developed across openPAT.
openPAT develops artificial intelligence technologies and software systems designed for different forms of interaction, reasoning, research, collaboration, and practical use.
Our work includes complete AI systems that bring together models, tools, interfaces, memory, research capabilities, and other supporting technologies into usable products and systems.
Normal Chat is the standard conversational experience within Lethan AI. It provides a direct interface for interacting with AI through everyday conversations, questions, problem solving, writing, and general tasks.
ALM (Action Language Model) is a technology developed as part of our work on AI interaction and action-based systems. It explores how AI can communicate and interact through structured actions rather than being limited to text responses.
TheTeam is a multi-agent system within Lethan AI where multiple AI agents can work together on a task.
Humanoid-LAE (Learn and Earn) explores how action-based AI can support a complete learning and assessment process — from first lesson to examination and certification. It is part of the Humanoid-LMT model family within Lethan AI.
Alongside products, openPAT continues to research and develop new AI technologies, software systems, and technical approaches.
Where to Go Next
Continue through the documentation to explore openPAT's products, technologies, research, and the people behind them.
Lethan AI — Explore openPAT's AI platform, its core capabilities, and the systems that work together to power the platform.
Humanoid-LAE — Learn about Learn and Earn, the action-based AI system for learning, examination, and certification within Lethan AI.
ALM — Learn about the Action Language Model and its approach to action-based AI interaction.
API — Explore APIs and developer interfaces for working with openPAT technologies and systems.
Research Updates — Explore ongoing research, technical developments, and areas being investigated across openPAT.
Organization
Information about how openPAT and its work are organized.
openPAT's work is organized across products, models, and technologies that come together across its AI ecosystem. Each area has its own purpose while remaining connected to the wider systems developed and used by openPAT.
Models
openPAT works with and develops artificial intelligence models across different areas of its technologies and products.
ALM
Humanoid-LMT 2 is an Action Language Model developed as part of LMT and focused on action-based AI interaction.
Humanoid-LAE (Learn and Earn) is part of the Humanoid-LMT model family, exploring learning, examination, and certification as an application of ALM technology within Lethan AI. It focuses on guiding learners from first lesson to final examination and certification.
LLMs
Orbital-LMT 3 is a large language model developed as part of the Lethan ecosystem.
Orbital-LMT 3 BLADE is a variant of Orbital-LMT 3, developed as part of the same model family and used within the Lethan ecosystem.
Technologies
Technologies form the underlying systems used across openPAT's products and AI work. These can include models, system components, interaction systems, behavioral systems, and other technical foundations.
LMT
LMT is a core technology developed by openPAT and the parent technology of ALM. It provides a broader technology layer for AI systems and capabilities within the Lethan ecosystem.
LCPE
LCPE (Lethan Construction Playground Engine) is a behavioral construction and feedback system within Lethan.
LCPE is based on CPE (Construction Playground Engine), which provides a system for turning verified feedback into structured behavioral constructions that can be stored, retrieved, and applied by AI systems.
Products
Lethan AI
Lethan AI is the main AI platform developed by openPAT. It brings together conversational AI, research, tools, memory, multi-agent systems, scheduling, document capabilities, and other technologies into a single platform.
Lethan incorporates technologies across the openPAT ecosystem, including LMT, ALM, LCPE, and Humanoid-LAE, allowing different systems and capabilities to work together within one platform.
Humanoid-LAE
Humanoid-LAE (Learn and Earn) explores how ALM technology could support a complete learning and assessment process — teaching, practice, examination, and certification — within Lethan AI.
Leadership
Siddhanth Jain is the Founder, CEO, and Developer of openPAT.
Research
Research and development work across openPAT's artificial intelligence technologies, systems, and products.
openPAT's research covers new AI technologies, model development, system design, and ways of making AI systems more capable and interactive. Research can lead to new technologies, product capabilities, or further areas of technical development.
TheTeam
Research and development on TheTeam was completed on September 22, 2026. The work focused on developing the underlying systems required for multiple AI agents to operate together as a coordinated team.
A major part focused on agent coordination, loop handling, and the Call Executor which handles execution and coordination of calls made during a TheTeam meeting.
Humanoid-LAE: Learn and Earn
openPAT is actively developing Humanoid-LAE (Learn and Earn), part of the Humanoid-LMT model family for Lethan AI. The work explores how Action Language Model technology could support a complete learning and assessment process — from first lesson to examination and certification.
The research covers AI teaching flows (teaching concepts, assigning exercises, examining submissions, identifying mistakes, adjusting lessons), examination systems (theoretical questions, code-writing, debugging, practical problem-solving), and certification infrastructure (unique IDs, permanent verification pages, subject and level recording).
Programming education (JavaScript, Python) is the initial focus, with potential to extend to other technical subjects.
ALM Models
openPAT is actively researching and developing ALM (Action Language Model) models. The work focuses on models designed for action-based AI interaction.
LCPE
Research is ongoing on LCPE (Lethan Construction Playground Engine). The work focuses on behavioral constructions, feedback verification, construction storage and retrieval.
Lethan
Lethan is an artificial intelligence platform that brings together conversation, research, models, tools, collaboration, scheduling, and interactive AI experiences in one system.
Lethan is designed to support different ways of working with AI, from everyday conversations and research to multi-agent collaboration and interactive work inside a digital workspace.
Models
Lethan uses different AI models across its capabilities, with models selected according to the type of work being performed.
Orbital-LMT 3
Orbital-LMT 3 is a large language model used within the Lethan ecosystem for general AI interactions and tasks.
Orbital-LMT 3 BLADE
Orbital-LMT 3 BLADE is a variant of Orbital-LMT 3 used within the Lethan ecosystem.
Humanoid-LMT 2
Humanoid-LMT 2 is the Action Language Model used by Lethan's interactive AI experience, Perspective.
Humanoid-LAE
Humanoid-LAE (Learn and Earn) is part of the Humanoid-LMT model family, focused on learning, examination, and certification as an application of action-based AI within Lethan.
Normal Chat
Normal Chat is the standard conversational experience in Lethan.
Perspective
Perspective is Lethan's interactive AI experience powered by ALM (Action Language Model).
Workspace
The Perspective workspace gives the AI a digital environment where it can work visually.
TheTeam
TheTeam is Lethan's multi-agent system, previously referred to as the Agent Council.
Research
Lethan provides research capabilities for tasks that require information to be gathered and examined. Deep Research and News Research.
Tools
Lethan can use a range of tools to perform tasks beyond conversation.
Scheduling
Lethan can schedule work to run automatically at a specified time.
Documents
Lethan can create and work with documents as part of its workflows.
Model Sources
Lethan does not rely exclusively on models developed within the Lethan ecosystem.
Humanoid-LAE: Bringing Learning, Examination, and Certification Into Action-Based AI
What if learning programming with AI went beyond asking questions and receiving answers? What if an AI could teach a student JavaScript, examine the code they write, identify their mistakes, test their understanding, and ultimately issue a certificate based on their examination performance? This is the idea behind Humanoid-LAE (Learn and Earn), part of the Humanoid-LMT model family being developed by openPAT for its AI platform, Lethan AI.
Humanoid-LAE explores how Action Language Model (ALM) technology could support a complete learning and assessment process, from a learner's first lesson to a final examination.
Understanding Humanoid-LAE
Traditional AI assistants can explain programming concepts and generate example code. However, a complete learning experience requires more than explanations. It requires a system that can follow a learner's progress, evaluate their understanding, identify knowledge gaps, and determine whether they can apply what they have learned.
Humanoid-LAE is designed around this broader objective.
Imagine a learner who wants to master JavaScript. Rather than receiving a collection of lessons and being left to study independently, the learner could interact with an AI teacher throughout the learning process.
The AI would teach concepts, ask questions, assign programming exercises, evaluate submitted work, and adjust subsequent lessons according to the learner's performance. The process would continue until the learner reaches the examination stage.
The focus is not simply on delivering educational information. It is on guiding the learner through a structured process in which knowledge must eventually be demonstrated.
How the AI Teacher Would Work
Consider a student learning JavaScript from the beginning.
The AI starts by teaching variables and data types. Once the student has studied the concept, it asks them to write a small program.
Suppose the student submits code containing an error.
The AI examines the submission, identifies the mistake, and explains why the code does not behave as expected. It can then ask the student to correct the problem or solve a similar exercise.
If the student continues to struggle with the same concept, the AI can revisit the explanation and provide additional practice.
As the student progresses, the lessons become more advanced, covering functions, arrays, objects, asynchronous programming, and more complex programming problems.
The same learning approach could be applied to Python and other subjects.
Active participation is central: Writing code, answering questions, correcting mistakes, and solving problems become part of the learning process rather than optional activities outside the classroom.
Examination: Testing What the Learner Actually Knows
Humanoid-LAE is also envisioned as an examination system.
After completing the required learning stages, a student would take assessments designed to evaluate their understanding of the subject.
For programming, examinations could include theoretical questions, code-writing challenges, debugging exercises, and practical problem-solving tasks.
The system would evaluate the student's answers and submitted code against defined assessment criteria.
For example, a JavaScript examination could test whether the student understands functions, can identify logical errors, and can write working solutions to programming problems.
The AI could also ask follow-up questions to investigate whether the student understands the reasoning behind their solutions.
This matters because producing a correct answer does not necessarily mean that someone understands the underlying concept.
The goal is to assess demonstrated knowledge rather than merely record lesson completion.
From Learning to Certification
The final stage of the Humanoid-LAE concept is certification.
After completing the required examination, a learner could receive an official certificate issued by openPAT, recording their assessed subject, level, and examination percentage.
For example, a learner who completes an advanced JavaScript assessment might receive a certificate recording their final result.
The proposed system could give each certificate a unique identification number and a permanent verification page, allowing others to check its authenticity in the future.
Certificates could cover different subjects and levels, from introductory programming to advanced technical knowledge.
The intention is to create a record of assessed achievement that learners can retain and share.
The credibility of these certificates would depend on the quality of the examinations, the reliability of the grading process, and the standards established by openPAT. Certification would represent performance against those standards, not an automatic guarantee of professional expertise.
The Role of ALM in Humanoid-LAE
Humanoid-LAE is connected to openPAT's work on Action Language Models (ALM).
The central idea behind action-based AI is to move beyond generating responses and towards systems that can carry out tasks within a digital environment.
Applied to education, this approach means coordinating a sequence of activities towards a defined objective: helping someone learn a subject and demonstrate their understanding.
That process can involve presenting a lesson, generating a relevant exercise, examining a submission, identifying an error, selecting the next learning activity, conducting an assessment, and preparing a result.
These activities are connected. Each step contributes to the learner's overall progress.
This is the perspective behind Humanoid-LAE: learning is treated as a process to be carried through, rather than a series of disconnected conversations.
Humanoid-LAE Within Lethan AI
Lethan AI is the AI platform developed by openPAT, bringing together conversational capabilities, research, and action-based digital work.
Humanoid-LAE belongs to the broader Humanoid-LMT model family associated with this development.
Its learning and examination concept represents one potential application of action-based AI: a system that can guide users through structured tasks, evaluate their work, and help them reach a defined outcome.
Programming education is one example. The same general approach could extend to other technical subjects and learning paths, provided that each has appropriate teaching material, assessment methods, and evaluation standards.
As Lethan AI approaches its launch, Humanoid-LAE represents an educational direction for openPAT's broader ALM research.
Conclusion
Humanoid-LAE (Learn and Earn) explores a model of AI-assisted education in which teaching, practice, examination, and certification are connected within a single learning process.
Its objective is to help learners move beyond receiving explanations to applying their knowledge, correcting mistakes, and demonstrating what they have learned.
For openPAT, this is an application of a broader idea: AI should be able to contribute to completing meaningful tasks, not just generating text about them.
LCPE
LCPE (Lethan Construction Playground Engine) is a behavioral construction and feedback system developed for Lethan. It provides a structured way for feedback about AI behavior to be reviewed, converted into constructions, stored, and later used during relevant interactions.
Instead of treating every piece of behavioral feedback as a temporary instruction, LCPE allows intended behavior to become a reusable construction. These constructions can describe how an AI system should respond in a particular situation and can be retrieved when a similar situation occurs in the future.
LCPE connects the full process from feedback to future AI behavior: feedback is submitted, verified, structured into a construction, stored in the construction system, and retrieved when it is relevant to an interaction.
Construction
A construction is a structured representation of a behavior that an AI system should follow in a particular situation.
Human Construction
Human construction allows feedback to be reviewed and turned into a construction directly by a human.
AI-Assisted Construction
LCPE can also use AI models to assist with the verification and structuring of submitted feedback.
Runtime
LCPE is not limited to the process of creating constructions. It can also participate when an AI interaction is taking place.
Verification Capacity
AI-assisted verification depends on the available capacity of the model used for the verification process.
CPE
LCPE is based on CPE (Construction Playground Engine), which provides the underlying system for creating, verifying, storing, and retrieving behavioral constructions.
API
CPE's construction database and retrieval capabilities can be exposed through an API.
Purpose
LCPE is designed to create a continuous connection between human feedback and future AI behavior.
Getting Started
Start working with openPAT technologies and developer resources.
openPAT provides developers with access to technologies, models, and APIs designed to be integrated into applications and other AI systems.
ALM API
The ALM (Action Language Model) API will provide developers with a way to integrate action-based AI capabilities into their own applications and systems.
Training Contexts
ALM models require large amounts of training context to learn how to operate across actions, environments, and interactions.
Lowering API Costs
By reducing the cost associated with the large training contexts required by action-based models, the technology can make the resulting ALM models more affordable to provide through an API.
Integration
Developers will interact with ALM through the API rather than managing the model's training process or training contexts themselves.
APIs
Programmatic interfaces for working with openPAT technologies and infrastructure.
openPAT provides APIs that allow developers to integrate selected technologies into their own applications and systems.
ALM API
The ALM (Action Language Model) API will allow developers to integrate action-based AI capabilities into their own applications.
TheTeam API
The TheTeam API will allow developers to integrate Lethan's multi-agent collaboration capabilities into their own applications and systems.
CPE API
The CPE (Construction Playground Engine) API will provide access to CPE's construction and retrieval capabilities for external AI systems and applications.
Humanoid-LAE API (Planned)
Future APIs are planned for Humanoid-LAE to enable learning, examination, and verifiable certification workflows in external applications.
API Availability
API availability may differ between technologies. Documentation for each available API will include authentication, request formats, responses, supported capabilities, and usage requirements.
Licensing
Licensing information for open-source projects and software released by openPAT.
openPAT releases certain software, source code, and technical work under open-source licenses.
Project Licenses
Each open-source project may use its own license depending on the nature of the software and how it is intended to be used.
Copyright
Unless stated otherwise, copyright in open-source work released by openPAT remains with the respective copyright holder.
Third-Party Software
Some openPAT software may use third-party libraries, frameworks, models, or other components that are distributed under their own licenses.
License Changes
Licensing may differ between versions or releases of a project.
Changelog
A record of notable changes across openPAT documentation and work.
Humanoid-LAE Documentation Added
Added comprehensive documentation for Humanoid-LAE (Learn and Earn), part of the Humanoid-LMT model family. Covers learning flows, AI teacher examination, certification with unique IDs and verification pages, and its role as an ALM application within Lethan AI.
Updated Organization page to include Humanoid-LAE, Research timeline to add active Humanoid-LAE research, and Products index to include Humanoid-LAE. Also added planned Humanoid-LAE API section.
Documentation Launch
The openPAT documentation was introduced as a central place for information about openPAT, its products, technologies, research, developer resources, and other technical work.
The initial documentation includes information about Lethan, LMT, ALM, TheTeam, LCPE, APIs, and open-source licensing.
Releases
Release information for products, technologies, and software developed by openPAT.
Humanoid-LAE — Learn and Earn
Documentation for Humanoid-LAE (Learn and Earn) published. Humanoid-LAE explores action-based AI for complete learning and assessment: teaching JavaScript/Python, examining code submissions, testing understanding through theoretical and practical assessments, and issuing verifiable certificates with unique IDs.
Part of Humanoid-LMT model family for Lethan AI. Programming education is initial focus, extendable to other technical subjects.
openPAT Documentation
The first public version of the openPAT documentation was published, bringing together information about the organization, products, technologies, research, developer resources, and licensing.
Announcements
Official updates and announcements from openPAT.
Introducing Humanoid-LAE: Learning, Examination, and Certification with Action-Based AI
openPAT introduces Humanoid-LAE (Learn and Earn), exploring a model of AI-assisted education in which teaching, practice, examination, and certification are connected within a single learning process. As part of Humanoid-LMT family for Lethan AI, Humanoid-LAE moves beyond Q&A to guide learners through structured tasks, evaluate work, and help reach defined outcomes with verifiable certificates.
Documentation now available under Products > Humanoid-LAE.
openPAT Documentation Is Now Available
The official openPAT documentation is now available, providing a central reference for openPAT's products, technologies, research, developer resources, and open-source work.
Security
Information about security across openPAT's products, systems, and technologies.
Security is considered throughout the development of openPAT's software, AI systems, infrastructure, and supporting services.
Privacy
Privacy-related information for openPAT products, services, and documentation.
openPAT considers privacy and responsible handling of information as part of the development of its products and services.
Legal
General legal information relating to openPAT and its products, technologies, and documented work.
openPAT develops and maintains software, artificial intelligence technologies, systems, documentation, and other technical work.
Terms
Terms and conditions relating to the use of openPAT services and resources.
openPAT products and services are currently under development.
Contact
Contact information for openPAT and its products, technologies, and documentation.
For questions about openPAT, its products, technologies, documentation, partnerships, or other matters, you can contact the founder directly.
Siddhanth Jain
Founder, CEO, and Developer of openPAT
Siddhanth Jain
Founder, CEO, and Developer of openPAT.
Siddhanth Jain is the founder, CEO, and developer of openPAT, a company focused on artificial intelligence, software, and the development of its own technologies and systems.
Based in Dharwad, Gandhinagar, Karnataka, India, Siddhanth works directly across the technical and creative side of openPAT.
Building openPAT
Siddhanth founded openPAT to develop and work on artificial intelligence and software technologies under one company.
He is directly involved in the development of Lethan AI, along with technologies including LMT, ALM, and LCPE, and now Humanoid-LAE.
Technical Work
Siddhanth develops and works on software across the openPAT ecosystem.
Role at openPAT
As Founder and CEO, Siddhanth leads the overall direction of openPAT and its products.