Systems for human-compatible intelligence

Bayon AI develops products and protocols that model context, roles, meaning, and interaction. This portfolio focuses on deployed applications, shared AI operations infrastructure, Web3 protocol work, and the research systems that inform future products.

AI Applications

Public product surfaces where Bayon AI turns specialized knowledge, relationship context, and domain workflows into usable software.

Specific wisdom through AI

Live

Primary

Guru AI

Guru AI is being built to help people hone in on specific methods for coaching, therapy, and learning through AI.

Primary consumer-facing AI product surface for turning specific wisdom into guided, interactive practice.

Visit Guru AI

Problem

Generic AI chat often stays broad. People need systems that can focus on a particular wisdom tradition, therapeutic approach, coaching method, or learning path and make that method usable in conversation.

Evidence

Live Guru AI web app

Persona and method-oriented chat direction

Account-based AI experience

Capabilities

AI-guided coaching and learning conversations
Character and method configuration
User accounts and continuity
Billing-ready product foundation

Stack

Next.js

Firebase

Stripe

Material UI

Relationship communication

Live

Active product

Bounded

Bounded facilitates communication in different kinds of relationships through information sharing between friends, colleagues, partners, and AI.

A relationship-centered product surface for privacy-aware sharing, reflection, and coordinated understanding.

Visit Bounded

Problem

Important relationship context is often scattered, private, or hard to share safely. Bounded gives people a structured place to capture, reflect on, and selectively share information with the people and AI systems involved.

Evidence

Live Bounded app

Private and shared entry model

Relationship-oriented information flows

Capabilities

Private and shared entries
Selective relationship context sharing
Friend, colleague, partner, and AI collaboration patterns
Flexible identity and entity modeling

Stack

Next.js

Firebase

Firestore

Material UI

Automated yoga teacher

Live

Active product

Yoga App

Yoga App is an automated yoga teacher that structures your practice on demand.

A domain-specific AI product that turns practice structure into a responsive, guided experience.

Visit Yoga App

Problem

Yoga practice needs sequencing, timing, guidance, and adaptation to the practitioner in the moment. A useful yoga app should organize the practice itself, not only list poses.

Evidence

Live yoga practice app

Sequence builder and practice flows

Responsive practice interface

Capabilities

On-demand yoga practice structure
Sequence building and practice timing
AI-assisted yoga generation
Responsive practice screens

Stack

Meteor

React

Cordova Android

Vercel

Event intelligence

Live

Active product

Akasha

An event intelligence platform for collecting, previewing, filtering, and managing public event data from multiple sources.

A vertical workflow candidate for turning messy public information into structured operational data.

Visit Akasha

Problem

Communities need reliable ways to gather fragmented public event information and turn it into searchable, manageable operational knowledge.

Evidence

Live Akasha deployment

Event ingestion workflow

Crawler and preview system

Capabilities

Public event source ingestion
Live scraping preview
Event filtering dashboard
Background sync and notifications

Stack

Next.js

Neon Postgres

Vercel

Tailwind

AI Operations

Shared operational systems that support assistants, agent creation, billing, usage visibility, and deployed AI product management.

Embeddable AI interface

Platform

Core platform

MiniChat Widget

A lightweight embeddable chat widget for adding an AI assistant or website concierge to external websites.

Reusable install surface for Bayon AI assistant deployments and AI website concierge services.

Visit MiniChat

Problem

AI and support conversations should be easy to add to existing sites without forcing teams to rebuild their frontend.

Evidence

Live widget deployment

One-line embed path

Website concierge fit

Capabilities

One-line embed
Theme and position customization
Persistent sessions
Accessible keyboard-friendly widget UI

Stack

Vanilla JavaScript

Webpack

Vercel

Iframe embedding

AI operations core

Platform

Core platform

AI Management Platform

The central core for AI operations, statistics, agent creation and management, billing, and shared platform controls.

Operational center for managing Bayon AI assistants, agent configuration, platform visibility, and commercial controls.

Visit AI Management Platform

Problem

A portfolio of AI applications needs one operational core for agents, usage, billing, and administration so product work does not become scattered across one-off dashboards.

Evidence

Live management platform

Agent and operations workflows

Billing and usage control surface

Capabilities

AI operations management
Agent creation and configuration
Usage and billing administration
Common operations controls

Stack

Next.js

Firebase

Stripe

Vercel

Web3

Protocol work for programmable identity, roles, groups, incentives, and composable social structure.

Composable social protocol

Web3

Research asset

SoulSystem Protocol

A protocol for representing users, groups, roles, incentives, and interaction flows through composable social primitives.

Protocol IP that informs identity, role, and social-structure thinking across human-compatible systems.

Visit SoulSystem

Problem

Online groups need programmable structures for identity, membership, role assignment, incentives, and governance.

Evidence

Protocol demo

Identity and role primitives

Composable procedure architecture

Capabilities

Soulbound profile primitives
Role and group representation
Incentive and reputation flows
Composable procedure design

Stack

Solidity

Hardhat

OpenZeppelin

TypeScript

Research

Long-running cognition and semantic systems that inform future human-compatible AI products.

Cognitive architecture

Research

Research asset

The GRID

A long-running cognition engine for modeling contexts, procedures, planning, and semantic state transitions.

Research foundation for representing context, procedures, and semantic behavior in future AI systems.

Problem

Human-compatible systems need more than chat completion. They need procedural memory, conditions, context, and ways to represent why an action should happen.

Evidence

Cognition engine IP

Procedure and condition modeling

Context tree experiments

Capabilities

Condition and procedure engine
Context tree modeling
Planning and event-triggered behavior
Client-side cognition experiments

Stack

Meteor

React

MongoDB

TypeScript

Building something that needs to understand people?

Bayon AI works across research, product design, and production engineering for systems that need context-aware interaction.

Talk to Bayon AI