// GLOSSARY

Kompozy Glossary

Plain-English definitions for AI content, autonomous publishing, and every term specific to the Kompozy stack.

Last verified · 2026-05-29 · by Moe Ameen

Square video aspect ratio (1080×1080) — the legacy Instagram feed format, still useful for cross-posting across many platforms.

Horizontal widescreen aspect ratio (1920×1080) — the standard for YouTube, webinars, and desktop-first viewing.

Portrait video aspect ratio (1080×1350) — the current preferred format for Instagram and Facebook feed (non-Reels) posts.

4D splat format

A volumetric video format that stores a moving scene as a cloud of time-aware Gaussian "splats" — letting a viewer move the camera freely through recorded motion instead of watching a fixed 2D frame.

Vertical video aspect ratio (1080×1920) — the native format for Reels, Shorts, TikTok, and Stories.

The repeating perceive-reason-act-observe cycle that turns a language model from a one-shot text generator into an agent that pursues a goal across multiple steps.

AI content farm

An AI content farm is a network that mass-produces low-quality, often inaccurate AI content to harvest ad revenue or spread propaganda.

AI content watermark

An umbrella term for any signal that marks content as AI-generated: a visible badge, an invisible SynthID mark, a C2PA metadata record, or a text watermark.

Using AI to translate and recreate a video's spoken audio in another language — usually in the speaker's own cloned voice, often with re-synced lips.

AI glossary (2026)

A plain-English reference to the AI terms creators actually run into in 2026 — LLM, token, prompt, hallucination, multimodal, agent, RAG, diffusion, fine-tuning, and inference — with what each one means for the person making content.

A technique that uses AI to reshape the mouth of a face in video so it matches an audio track — used to dub, re-voice, or animate a talking face without reshooting.

AI rage-bait is provocative content mass-produced by AI — fake celebrity clips, staged outrage — to farm anger, engagement, and reach at near-zero cost.

Low-quality, generic media mass-produced by generative AI with little human oversight, and now the content audiences and platforms increasingly reject.

AI text watermarking

A hidden statistical signal embedded in an AI model’s word choices as it writes, letting a detector later confirm the text was machine-generated.

AI-generated, deliberately alluring photo or video content from a synthetic persona, engineered to bait clicks, follows, and paid subscriptions.

AI-generated video engineered to look like a real person casually filming a phone testimonial: the user-generated-content style, produced without any creator.

How often AI answer engines like ChatGPT, Perplexity, and Google's AI Overviews mention, cite, or recommend your brand — the measurable outcome that GEO moves.

A scam that uses a synthetic clone of a real person's voice — often built from as little as three seconds of public audio — to impersonate them over a phone call or voice message and extract money or access.

AI voice generation

Turning written text into natural, human-sounding speech with a neural model — used to voice videos, podcasts, and narration without a recording session.

The ranking and distribution system a platform uses to decide which content gets shown to which users, in what order.

The proportional relationship of a video or image’s width to its height, expressed as W:H (e.g. 9:16, 1:1, 16:9).

Atomized content

A long-form source broken down into the smallest standalone units of value — each able to be published independently.

A short audio clip from a podcast or interview, paired with a waveform animation, captions, and a still image — designed for social-feed sharing.

Kompozy’s opt-in mode that generates and schedules content without human approval — gated by 4 quality checks.

AI-generated talking-head video where a digital avatar speaks a written script using voice cloning or synthetic voice.

Supplementary footage layered over the main shot (A-roll) to illustrate a point, hide cuts, or maintain visual interest.

Brand messaging

The strategic set of core messages — value proposition, positioning, message pillars, and tone — a brand repeats consistently across every channel.

The text body that accompanies a social-media post — on Instagram and LinkedIn, often the difference between scroll-past and engagement.

A vertical short-form video cut from a longer source (podcast, webinar, YouTube long-form) with auto-captions.

Content anchoring

Content anchoring is organizing every post around one central asset, theme, or buyer belief, so each piece reinforces the same idea instead of drifting.

A near-synonym for content pillar, often used to describe a recurring content format (e.g. "Tip Tuesday", "Case Study Friday") rather than topic.

Content calendar

A scheduled plan that maps content pieces to publish dates, platforms, and themes — the operational backbone of consistent posting.

Content flywheel

A system where each piece of content produces inputs (audience, feedback, ideas) for the next piece — compounding output over time.

Content pillars

The 3–5 core themes or topic categories every piece of content for an account maps to — the editorial spine of a content strategy.

Content repurposing

Converting one piece of source content (podcast, video, blog) into multiple output formats across multiple platforms.

Context Engineering

Curating the whole set of tokens a model sees at inference — instructions, tools, references, memory, retrieved data — not just the wording of one prompt.

Credit-based pricing

A pricing model where each output costs a fixed number of credits tied to its actual cost-to-serve — replacing per-seat pricing for AI tools.

Publishing the same content asset to multiple platforms — efficient at scale but penalized when done without per-platform formatting.

Call-to-action — an explicit instruction telling the viewer what to do next (follow, comment, click, subscribe, save).

Click-through rate — the percentage of impressions that result in a click. On YouTube, the percent who clicked after seeing the thumbnail.

Diffusion language model

A language model that generates text by denoising a whole sequence in parallel — like an image model — instead of predicting one token at a time.

Engagement rate

The percentage of viewers who took an action (like, comment, share, save) divided by total reach or impressions.

A uniquely identifiable thing — a person, brand, place, or concept — that a search engine can pin to one distinct record in its knowledge graph.

Evergreen content

Content that stays relevant and continues to generate views, traffic, or leads months or years after publishing — the opposite of news.

Faceless AI video generator

A tool that produces finished video without the creator on camera — AI voiceover, generated or stock visuals, captions, and often auto-posting.

Generative Engine Optimization (GEO)

The practice of shaping content so AI answer engines like ChatGPT, Perplexity, and Google’s AI Overviews cite and quote it in their generated answers.

A keyword prefixed with # that categorizes content and (on some platforms) drives discovery via hashtag search and follow.

The opening 1–3 seconds of a video or first line of a post — designed to stop the scroll and earn the next 5 seconds of attention.

Horizontal video

Video shot at 16:9 aspect ratio — the standard for YouTube long-form, webinars, and traditional broadcast.

Kompozy’s self-hosted HTML→MP4 template renderer that burns in branded captions and overlays pixel-exact to your style.

The total number of times a piece of content was displayed, including multiple views by the same person.

An edit that removes a chunk of the same continuous shot, producing a visible skip — used to tighten pacing in talking-head video.

Likeness detection

Platform technology that scans uploads for a specific enrolled person’s face or voice and flags AI-generated content using their identity, so they can review it or request removal.

A Markdown file at a site's root that hands AI agents a clean, linkable map of its key pages. The 2026 V2 update adds formal Markdown link relations.

Long-form video

Video longer than ~3 minutes, typically published on YouTube, podcasts, or as webinars — optimized for watch time and depth, not feed discovery.

Marketing Shorts

A 4-second HeyGen hook composited with demo footage and music for paid-ad creative.

The Paris-based AI lab known for open-weight large language models you can download and self-host — Europe’s leading answer to OpenAI and Anthropic.

Model Context Protocol (MCP)

An open standard that lets an AI assistant connect to your files, tools, and data sources through one common interface — the "USB-C port for AI" that marketers use to give a model live access to their analytics, CRM, and CMS.

Non-Impersonation Prompt

A system-prompt instruction, or set of them, that stops an AI from claiming or implying it is human — the core technique for honest AI tone and disclosure.

Omnichannel content

A distribution strategy where the same core message is reshaped for every platform an audience uses, instead of choosing one channel.

The five user-facing content categories every Kompozy generation maps to: Video, Image, Text, Blog, Newsletter.

Painting with Gaussians

An emerging technique that builds an image from thousands of 2D Gaussian "splats" used as brush strokes, a painterly alternative to pixel grids and diffusion.

A structured prompt that defines your voice, banned words, reference creators, and required formats — used as context for every AI-generated output in Kompozy.

A Kompozy render path that wraps a HeyGen avatar inside a HyperFrames composition template for branded scene shorts.

Kompozy’s default avatar-video path: HeyGen avatar plus auto-captions plus optional B-roll, without a HyperFrames template.

A tweet-styled image card rendered with your avatar, handle, and brand styling — postable anywhere, not just X.

Photo-to-Video AI

Photo-to-video AI turns a still image into a short animated clip, either by adding realistic motion to the scene or by making a portrait speak and lip-sync.

AI porn is sexually explicit images, video, or audio generated by AI, spanning consensual synthetic content and non-consensual deepfakes targeted by 2026 law.

Prompt injection as role confusion

A framing of prompt injection as a failure of role perception: LLMs identify who is speaking from how text sounds, not from its labeled role, so attacker text written in a trusted style inherits that trust.

Four automated checks every Kompozy output passes before autopilot ships it: persona, platform-cadence, fact-anchor, brand-safety.

The number of unique accounts that saw a piece of content at least once — distinct from impressions, which counts repeated views.

Real-time phone call captions

Live, on-screen text of what the other person is saying on a phone or video call, transcribed by streaming AI speech recognition as they speak — with sub-second latency, not after the call.

Retention curve

A graph showing the percentage of viewers still watching at each point in a video — the primary signal algorithms use to rank video.

Retrieval-augmented generation (RAG)

RAG (retrieval-augmented generation) retrieves relevant documents at query time and feeds them to an LLM, so answers are grounded in your sources, not guessed.

Schema markup for AI citations

Using schema.org structured data to make a page machine-readable and its entity verifiable for AI answer engines — an assist to citation, not a cause of it.

A reduction in a piece of content’s or account’s distribution without an explicit ban notification — the platform silently suppresses reach.

Short-form video

Vertical or square video typically under 60–90 seconds, optimized for feed scrolling and algorithmic discovery on Reels, Shorts, and TikTok.

Social media AI agent

Software that pursues a social-media goal on its own — planning, creating, scheduling, and replying — via an LLM with tools, memory, and a feedback loop.

Social Media MCP

A Model Context Protocol server that connects an AI agent to your social tools, so it can read your data and publish to your accounts on instruction.

Social media monitoring

Tracking what people say about your brand across social platforms in real time — tagged mentions and untagged conversation — so you can respond fast.

On-screen text transcribing spoken dialogue in a video — required for sound-off viewing on every modern social-video feed.

A system prompt is the foundational instruction block sent to an LLM before a conversation. It sets the model’s role, tone, rules, and priorities.

The AI Aesthetic

The design language of AI products themselves — the sparkle icon, beige-and-serif look, streaming and shimmer text — not the look of AI-generated content.

The AI Design Aesthetic

The recognizable visual style generative tools converge on by default — glossy, hyper-saturated, symmetrical, and uncannily smooth — now common enough that audiences and algorithms spot it on sight.

The static preview image that represents a video in feeds, search results, and channel pages — the single biggest driver of CTR.

User-generated content — content made by customers, audience members, or hired creators that looks unscripted and authentic rather than brand-produced.

Video shot or cropped at 9:16 aspect ratio, optimized for phone-held viewing on Reels, Shorts, TikTok, and Stories.

How a platform counts a 'video view' — from TikTok's instant, loop-counting play to the MRC 2-second standard used by X, LinkedIn, and Pinterest.

Viral clip detection

An algorithm that scans a long-form video and predicts which short segments are most likely to perform as standalone shorts.

Visible AI watermark

An on-file badge — like Gemini's corner sparkle or a 'Made with AI' label — that shows a viewer content is AI-generated, unlike an invisible SynthID mark.

Streaming or making video as an animated virtual avatar — a Live2D or 3D character driven in real time by a human performer's face, voice, and movement.

The total cumulative minutes viewers spent watching a video — YouTube’s primary ranking signal for long-form content.

A Kompozy container that holds exactly one Persona Brief, one asset library, and one set of source and outlet configurations.

Frequently asked questions

What is the Kompozy glossary?

The Kompozy glossary is a set of plain-English definitions for the terms used across the Kompozy stack — Persona Brief, HyperFrames, autopilot, credit-based pricing, avatar video, viral clip detection, and more. Each term links to a fuller definition page.

What is a Persona Brief?

A Persona Brief is Kompozy's brand-voice system — the codified fonts, colors, avatar, handle, and voice that every output renders against so the result is pixel-true to your brand rather than a generic stock look.

What are HyperFrames?

HyperFrames are Kompozy's pre-wired composition templates. Each one turns a source recording into a finished, brand-exact, platform-native render with captions, overlays, frames, and outro baked in — so you pick a proven format pattern instead of editing from scratch.

What does autopilot mean in Kompozy?

Autopilot is the mode where Kompozy ingests a source, composes the derivative outputs, runs them through quality gates, and publishes across platforms without manual review of each piece. It is the autonomous end of the workflow, as opposed to the review-dashboard path.

How does credit-based pricing work?

Credit-based pricing charges per output rendered rather than per seat. Different output types cost different credit amounts, and your plan includes a monthly credit allowance you spend across whichever formats you generate.

Where can I find the full definition for a term?

Every term in the glossary grid links to its own definition page with a longer explanation and related terms. Click any term card to open the full entry.