Mariah Assistant · Powered by the Kam-1 Engine

One Intelligence Layer Across Every Ploonu App.

Meet Mariah—Ploonu’s consumer AI assistant—and Kam, our hybrid multi-modal model engine uniting on-device Apple Intelligence with edge neural routing.

Try Mariah Interactive Demo Inspect Kam-1 Architecture

The Consumer Experience

Meet Mariah

Whether you are drafting a post in Yaps, building a 6-slide photo + audio story in Vidiloop, or publishing in Typepress, users interact with one unified creative partner: Mariah.

The Underlying Model Engine

Powered by Kam

Under the hood, every request routes through the Kam engine (spelled with a K). On iOS, Kam executes on-device via Apple Intelligence (FoundationModels) at $0 server cost, with edge fallback on Web/PWA.

Experience Mariah + Kam in Action

Select a live scenario below to see how Mariah routes tasks through the Kam engine across Yaps and Vidiloop.

api.mariahlabs.com/v1/kam · Apple FoundationModels + Edge

Assistant: Mariah (Yaps iOS Composer) Engine: kam-1 (On-Device Apple FoundationModels · 0ms Cloud Egress)

Drafting with Mariah: "Breaking: Game 7 goes to overtime! Who you got taking the series tonight? Drop your picks below 👇 #NBAPlayoffs" ✓ Tone tightened for high engagement ✓ Hashtag & @yapssports category auto-matched ✓ Executed locally on iPhone Neural Engine ($0.00 token burn)

Where Mariah Lives

Built once at Mariah Labs. Deployed natively across the entire Ploonu product portfolio.

Real-time social conversations, clubs, and breaking verticals at yaps.club.

Draft & Review with Mariah inside the Yap Composer accessory bar
On-device Apple Intelligence (FoundationModels) integration
Creator Analytics insights in Settings
Suggest-only curation for 60 official @club & @yaps... handles

Visit yaps.club →

Short-form video loops and 6-slide photo + audio stories at vidiloop.com.

kam-vision smart framing & 6-slide photo sequencing
kam-audio voiceover transcription & volume sync
Automated silent-loop detection & caption generation
Agnostic Publisher workflows for *sports verticals

Visit vidiloop.com →

Next-generation publishing, developer workflows, and prompt intelligence.

Typepress: Long-form editorial copilot powered by Mariah
Mergemate: Autonomous PR & conflict synthesis
Promptalyzer: Local kam-1 prompt scoring & benchmarking
Public Roadmap: Future open-weights release of the Kam model

View Kam Roadmap →

The Kam Model Matrix

Provider-agnostic, privacy-first routing. On-device first, edge second, zero wasted compute.

Model ID	Modality & Role	Primary Execution Runtime	Ploonu App Coverage
kam-1	Flagship Reasoning, Writing & Review	Apple `FoundationModels` (iOS) · Workers AI / Edge (Web)	Yaps Composer, Creator Analytics, Typepress
kam-fast	Sub-100ms Autocomplete, Tags & Triage	On-Device Neural Engine · Edge Llama 3.3	Yaps Category Pill, Hashtag Suggestions, Search
kam-vision	Photo Sequencing, Framing & Visual Captions	Apple Vision / CoreML · Multimodal Edge	Vidiloop 6-Slide Photo Studio, Yaps ComposePhotos
kam-audio	Voiceover Transcription & Audio Mastering	Faster-Whisper (`small.en`) · On-Device Speech	Vidiloop Audio Studio, Sentinel Voice Notes