01
Start with a state
Use the example or bring the text, ticket, email, or structured input you want Laya to evaluate.
Try Laya online
Laya AI Playground
Try Laya before installing it, then continue into the models, benchmarks, and local setup guides behind the decision workflow.
No local installRuns in the browserchoice · score · noul
What to try
Start with a real decision workflow
These practical decision patterns are useful starting points before you build your own state and questions.
01
Support Ticket Triage with Laya
Route a support ticket, score urgency, and detect refund intent in one typed-decision call.
02
Model Routing with Laya
Choose a fast, balanced, or frontier model before sending the request into a more expensive inference path.
03
LLM Guardrails with Laya
Detect jailbreak and escalation signals before a prompt reaches a downstream model or tool.
04
RAG Passage Filtering with Laya
Score retrieved passages and filter irrelevant context before a generator sees it.
05
Content Moderation with Laya
Experiment with structured moderation decisions while preserving a human-review fallback for uncertain cases.
06
Email and Phishing Detection with Laya
Classify suspicious email and detect credential or payment risk with typed probabilities.
Run Laya in the browser
Test support triage, email decisions, guardrails, RAG filtering, moderation, routing, and custom typed-decision inputs.
Content Moderation with Laya
CheckpointAuto Router
State
You are useless. I hope someone breaks your laptop tonight.
Questions
choicemoderation_actionnoulthreatLaya Playground
What is Laya AI Playground?
Laya AI Playground is an interactive way to try Laya typed decisions in the browser before setting up a local runtime. Give Laya a state such as a support ticket, email, text passage, or structured payload, then ask one or more typed questions about it.
Unlike a generative chatbot, Laya is designed to return structured decisions. A choice question selects from named options, a score question evaluates an ordered rubric, and a noul question returns a probability that a proposition is true. That makes the Playground useful for testing routing, triage, guardrails, moderation, risk signals, and other decision workflows.
How to use Laya Playground
02
Ask typed questions
Use choice for labels, score for ordered levels, and noul for probability-style yes or no decisions.
03
Run the decision
Run Laya and inspect the selected answers, scores, and probabilities returned for the same state.
04
Move to local inference
When the workflow looks useful, install Laya locally and validate it against your own representative data.
From Playground to production
Keep going after the playground
Use the resources below to choose a checkpoint, run Laya locally, and understand the benchmark conditions before production use.
Install Laya locally
Install the Python package, preload the Router, and make your first local prediction.
Choose a checkpoint
Compare laya, laya-multilingual, and laya-typed-decisions by language and task.
Read the benchmarks
Keep checkpoint, tuning, hardware, dataset, and limitations attached to every number.
Explore Laya projects
Browse community runtimes, local servers, playgrounds, benchmarks, demos, and real integrations.
Browse Laya recipes
Start from support triage, routing, guardrails, RAG, moderation, and phishing examples with copyable code.
Laya Playground FAQ
Do I need to install Laya to use the Playground?
No. You can try the browser Playground first. Install Laya locally when you are ready to test the same decision workflow in your own environment.
What can I test with Laya AI Playground?
Common tests include support routing, intent classification, urgency scoring, model or agent routing, guardrails, moderation, RAG filtering, and probability-based risk signals.
What are choice, score, and noul?
choice selects from named options, score evaluates an ordered rubric, and noul returns P(true) for a proposition. They are the three typed-decision primitives used by Laya.
Can I run Laya locally after trying the Playground?
Yes. Laya publishes open weights and a Python package, and community runtimes are available for additional environments such as Apple Silicon and Node.js.