◈
Conversational agents
Agents work with conversation history, prepared context and tool-call loops to handle requests.
JAVA-FIRST · ENTERPRISE AI
Intelligence,
orchestrated.
Meet AxiomUI — an AI-powered automation platform built in pure Java. Bring agents, tools, memory and persistent workflows together in one extensible system.
JDK 25+Agent orchestrationPersistent automation
01 / AGENT · context aware
02 / ROUTER · model selection
03 / TOOLS · action layer
01 — THE PLATFORM
More than a chat window.
AxiomUI connects conversational AI with the mechanisms needed to act: context, tools, scheduled work and operational controls.
⌁
Model router
Classifies prompts and routes them to the first usable configured model for that task class.
⌘
Tools & integrations
Extend agents with tools, MCP servers, reusable skills and messaging channels.
◷
Persistent jobs
Schedule recurring work with persisted execution, retries and crash recovery.
▤
Memory & context
Maintain conversations across sessions and prepare relevant context for agent runs.
⌬
Multimodal capabilities
Voice, image and video features are available through configured providers and optional sidecars.
⛨
Guarded actions
Dangerous actions can require approval, with optional OS sandboxing for supported processes.
◎
Operational visibility
Opt-in OpenTelemetry traces and metrics help inspect model calls and system activity.
02 — INTERACTIVE EXPERIENCE
Step inside the assistant.
Try a guided product simulation. Choose a scenario and run it to see how an agent workflow might appear to an end user.
AxiomUI / PLAYGROUNDLOCAL INTERACTION
✳AxiomUI AssistantDEMO
Hello. Choose a sample task below and run the simulation to explore the agent experience.
This embedded playground is a front-end simulation. It does not connect to an LLM, execute tools, or send external requests.
03 — UNDER THE HOOD
Follow the intelligence.
Select a component to explore its role in the conceptual request lifecycle.
◉ User & messaging channelsWeb · Telegram · Slack · WhatsApp
↓
✳ Agent orchestrationContext · routing · tool loop
↓
◇ Model routerTask class · configured providers
↓
⌘ Tools & MCPActions and integrations
▤ MemoryConversation context
◷ SchedulerPersistent tasks
⛨ SecurityApprovals & sandbox
ENTRY POINT
User & messaging channels
AxiomUI can receive conversations through its web interface and configured messaging-channel bindings. Incoming messages are dispatched into the agent system.
SELECT A NODE TO INSPECT
04 — ENGINEERING
Java first. Deliberately lean.
AxiomUI is built on a customized Play Framework 1.x foundation, using focused libraries and JVM capabilities rather than a heavy application framework.
Java 25+Primary application runtime
Play Framework 1.xCustomized application foundation
OkHttp 5HTTP client primitives
JPAPersistence layer
db-schedulerPersistent scheduled execution
Virtual threadsJVM concurrency capability
Nuxt 4 · Vue 3Built-in SPA frontend
TypeScriptFrontend language
01 / DESIGN
Focused primitives
Lean libraries support HTTP, persistence, scheduling and process execution.
02 / OPERATIONS
Upgrade workflow
The installer supports in-place upgrades, database backup and rollback on failed startup.
03 / EXTENSIBILITY
Modular capabilities
Agents can be extended with skills, MCP tools and configured channels.
04 / OBSERVABILITY
Traceable model calls
Opt-in OTLP tracing and metrics include GenAI spans.
05 — PROJECT SNAPSHOT
Built to be explored.
A source-available Java AI automation platform from Abundent. Review the implementation, documentation and license before adopting it.
PROJECT FACTS
25+
JDK runtime
Java
Primary server language
OTLP
Opt-in observability
SPA
Built-in frontend
EXPLORE THE SOURCE
Go beyond the interface.
Inspect the code, releases, setup instructions and licensing terms in the project's repository.
Licensing: source-available under PolyForm Noncommercial 1.0.0 for noncommercial use; commercial use requires a commercial license. Versions through v0.15.4 remain under MIT.