01
Signal-based clip detection
PeakCut scores moments from four signals — chat spikes, audio energy, on-screen events, and the story in the transcript — so the clips it surfaces are the ones that actually pop.

AI video automation platform
PeakCut is a full product delivery record for an AI video automation platform: it takes a long stream, interview, or raw footage and produces publish-ready short-form content without manual editing. The system scores peak moments from multiple signals, cuts vertical clips with karaoke captions, generates faceless story videos and AI-presenter episodes, speaks in a cloned voice across many languages, and schedules everything to YouTube Shorts, TikTok, Reels, and X. It proves Nythral can ship a complete AI product — pipeline, editor, generation, billing, and a launch-grade marketing site — as one system.


Delivery record
The work connected marketing, CRM, marketplace extensions, microservices, and daily shop operations into one product architecture.
01
PeakCut scores moments from four signals — chat spikes, audio energy, on-screen events, and the story in the transcript — so the clips it surfaces are the ones that actually pop.
02
Beyond clipping, the platform generates original content: AI presenters, faceless explainers, and synthesized voiceovers, rendered from a topic rather than filmed.
03
A timeline editor allows caption, music, and scene changes without re-rendering, and the output is packaged per platform with titles, hashtags, and scheduling.
04
The build includes subscription tiers, a generation credit system, onboarding and waitlist flow, transactional email on Nmail, and a public beta marketing site.
Product screens
Key screens, workflow states, launch pages, and operating surfaces are arranged as a moving system instead of static thumbnails.

Peak detection

AI renders

How it works

Content studio
Product system
AI video platform / clip automation / multi-platform publishing
Peak detection
AI video generation
Voice cloning
Timeline editor
Trend autopilot
Multi-platform publishing
Billing and credits
Multi-signal analysis turns hours of footage into a ranked shortlist of publish-worthy moments automatically.
AI presenters, faceless stories, and cloned-voice narration make it a content studio rather than a trimming tool.
Pipeline, editor, publishing, billing, and a launch site delivered as one shipped AI system.
Outcomes
01 / 04

Peak detection
The core promise on the landing surface: paste a long VOD and PeakCut scans it, scoring peak moments and surfacing the clips worth publishing.
01
Built an AI pipeline that scores peak moments from chat activity, audio energy, on-screen events, and transcript narrative to auto-select clips.
02
Delivered AI generation: faceless story videos, AI-presenter episodes, and cloned-voice narration across many languages.
03
Shipped a timeline-based editor for caption styling, music, and scene order plus per-platform titles, descriptions, hashtags, and scheduling.
04
Wrapped it with billing tiers, a credit system, onboarding, transactional email, and a launch-grade marketing site.
Technology stack
The stack was selected around the delivery goal, production constraints, operating model, and future iteration path.
marketing site and product web surfaces
typed product, pipeline, and service logic
backend runtime and job orchestration
accounts, jobs, credits, and content data
AI inference and video rendering direction
clip cutting, captions, and video composition
subscription tiers and generation credits
transactional email — signup, waitlist, and notifications
production hosting and delivery
Start a similar build
Use the same request flow as the main site. We will route the conversation to product strategy, mobile app, CRM/admin, agentic engineering, or launch-site work.