A Laya checkpoint trained on PolyAI's Banking77, the 77-intent banking support dataset, tagged for customer-support intent classification.
GitHubSupport & triage
Support & triage
6 builds
Homebrew menu bar server that turns n8n Text Classifier requests into Laya typed questions in ~40 ms. Weights ship inside the app; no network, unloads when idle.
GitHubSupport & triage
Laya Serve for macOS: a menu bar app with an OpenAI-compatible endpoint for n8n
- Memory loaded
- ~1.3 GB
Laya shipped right after Jev claiming to be faster for AI agent decisions. curious if it was accurate too I built a POC same Google ADK triage agent, same 30 labeled tickets, only the engine swapped. Jev: 422ms, 53% accuracy. Laya: 152ms 10% accuracy Faster not more accurate
X postSupport & triage
Same Google ADK triage agent, engine swapped: Jev 53% at 422 ms, Laya 10% at 152 ms
- Accuracy
- 10% vs 53%
- Tickets
- 30
Spent a morning testing a small open-weights local model (Laya-MLX, fully offline on a MacBook Air, $0 per call) on real tasks: inbox triage, tagging 100 SAP decks, categorizing blog posts, and a prefilter pattern. Honest numbers in this thread.
X postSupport & triage
A morning of real tasks on Laya-MLX: inbox triage, 100 SAP decks, blog categories
1,000 synthetic emails. Six folders. One local 322M Laya model running on a MacBook Air M3. ⚡ 28.6s without screen recording 🎬 97.3s while recording 🎯 65.1% accuracy against reference labels 💾 0 MB swap Everything runs locally. No cloud inference. And I’m showing the
X postSupport & triage
1,000 synthetic emails into six folders with local Laya on a MacBook Air
- Accuracy
- 65.1%
- Swap
- 0 MB
Gradio demo of Laya on Hugging Face: email triage, phishing, guardrails, ticket routing, RAG filtering and multilingual routing, no install.
SiteSupport & triage
Laya demo Space: eight workflows in the browser
- Spam accuracy
- 99.3%
- Phishing accuracy
- 98%