Featured
sysd
Local-first distributed systems workbench combining Excalidraw scene-graph diagnostics, 11 Staff+ reference blueprints, Free-Flow practice, Cactus Needle 3 + Whistle WASM, and Kokoro TTS.
WASM · Excalidraw · Kokoro TTS
Ninada (निनाद)
Sub-second in-browser 24 kHz streaming TTS across Classical Sanskrit, English, and 10 Indian languages via 48 capability-routed HiFi-GAN & iSTFTNet ONNX voices.
ONNX · WebGPU & WASM · 12 Languages
Swanada (स्वनद)
Sub-second in-browser 16 kHz streaming speech-to-text across Classical Sanskrit, Indian English, and 10 Indian languages via shared Conformer INT8 + 515 KB CTC heads.
ONNX · WASM SIMD · 12 Languages
XKCD-Bench
Blind multimodal humor & reverse comic search benchmark across 32 frontier LLMs (Gemini 4 Carbon, Muse Wonder, Kimi K3.1 Pro, Claude Fable 5.2, GPT-6 Astra Pro, Grok 4.7) and 891 strips.
Eval · Gemini 4 Carbon · TypeSafe
DDx · Differential
Science-baked clinical differential diagnosis & bedside triage across 49 ICD-10 pathologies via QMR-DT Bayesian networks and OpenJev Qwen3.5-4B WASM.
WASM · QMR-DT · OpenJev
Eloquent Web
On-device AI dictation powered by Gemma 4 and WebGPU. Speak naturally, get polished text.
WebGPU · Gemma 4
Pocket Voice
Clone any voice in your browser. Record, clone, and speak, 100% client-side via ONNX.
ONNX · Voice Cloning
Visual Narrator
Visual storytelling powered by Qwen 3.5 0.8B running via WebGPU.
WebGPU · Qwen 3.5
EM Tools
Engineering management toolkit for team leads and managers.
Productivity
PM Tools
AI utilities for specs, PRDs, and product planning.
Productivity
Voice & Language
Ninada (निनाद)
Sub-second 24 kHz streaming text-to-speech across Sanskrit, English, and 10 Indian languages via 21.7 MB iSTFTNet & 27.9 MB HiFi-GAN ONNX voices.
Swanada (स्वनद)
Sub-second 16 kHz streaming speech-to-text across Sanskrit, Indian English, and 10 Indian languages with ANC + Wiener noise cancellation and 0% prefix churn.
Vāgdhenu (वाग्धेनु)
100% in-browser Sanskrit śloka-to-chant synthesis via ONNX Runtime (WebGPU & WASM) with Pāṇinian Chandas scansion.
Su-śrotā (सुश्रोता)
On-device Sanskrit speech recognition via Conformer-CTC with live audio and IAST transliteration.
Pocket Voice
Clone any voice in your browser. Record, clone, and speak, 100% client-side.
Voice Summarizer
Record or upload audio and get an AI-generated summary.
Nadha
AI music generation and sound exploration tool.
TextCraft
Transform, reformat, and polish text with AI assistance.
Debate
AI-powered debate simulator for exploring both sides of an argument.
Lipi (ಲಿಪಿ)
Offline voice-to-script studio for Kannada, Sanskrit, Hindi, Tamil, and Telugu powered by Whisper WASM.
Phonica
Interactive phonics and pronunciation guide for children learning to read aloud.
Chrome Built-in AI
Knowledge & Research
DDx · Differential
Bedside clinical differential diagnosis, ROS charting (+ / −), HEART/Wells/CURB-65 calculators, and SOAP/MDM note studio across 49 ICD-10 pathologies.
Explain Paper
Paste an academic paper and get a plain-language explanation.
Book Summary
Get concise summaries and key takeaways from any book.
First Principles
Break down any concept into its fundamental building blocks.
FollowUp
Generate smart follow-up questions for deeper exploration.
Prompt Tuner
Refine and optimize prompts for better AI outputs.
PM Tools
AI utilities for specs, PRDs, and product planning.
Creative & Visual
Culture & Knowledge
Developer
sysd
Distributed systems architecture studio with live Excalidraw topology lint, 11 Staff-level blueprints, Free-Flow practice, and on-device neural voice coaching.
XKCD-Bench
Multimodal LLM comic explanation arena & calibrated reverse comic search benchmark across 891 Hugging Face XKCD strips.
FunctionGemma
Function-calling with Gemma models via WebGPU.
LLM Inference
Run large language models directly in the browser.
Models
Browse and compare available on-device AI models.
Browser LLM Fit
Hardware capability analyzer to check which LLMs fit your browser's WebGPU VRAM.
EM Tools
Engineering management tools for team leads.