# AI Content Drop

> AI Content Drop is an AI-powered video, image, and content generation platform at https://aicontentdrop.com. Its thesis: winning ads and viral videos aren't guessed — they're reverse-engineered. An agent studies the top ads in your market, distills each winner into a reusable Skill (hook, script, scenes, sound, CTA), and generates your version in your brand voice across 45+ frontier models.

## What AI Content Drop Does

AI Content Drop is a unified platform for creating AI-generated video ads, UGC content, images, and marketing materials. It is not a single-model tool — it aggregates 45+ AI video and image models from the leading providers (Kling, Veo, Seedance, Wan, Hailuo, Grok, and more) into one interface with a single credit system.

The platform's core loop is the Winning-Ad Engine: Scout (the agent sweeps public ad libraries and viral channels for the ads your market keeps paying to run) → Decode (every winner is distilled into a reusable Skill — hook, script skeleton, scene rhythm, audio profile, CTA pattern — with proof attached, like how long the ad has been running) → Reshoot (your product, your brand voice, generated as entirely new footage). The platform extracts the formula, not the footage.

The generation interface is the Chat-to-Ads Studio: users describe their video ad in natural language, and the platform selects the optimal AI model, generates the content, and delivers it. This is powered by a multi-LLM routing system that classifies user intent and dispatches to specialized providers.

## Key Features

- Chat-to-Ads Studio: Conversational AI ad creation with multi-LLM routing (Gemini, Claude, GPT)
- AI Video Generation: 45+ video and image models including Kling 3.0, Veo 3, Seedance, Wan 2.5, Hailuo, MiniMax H3
- AI Image Generation: 10+ models including GPT Image 2.5 Flare and Sunburst (OpenAI, September 2026), Nano Banana Pro, Grok Imagine, Z-Image
- UGC Avatar Factory: AI-generated spokesperson videos with lip-sync and natural text-to-speech voiceover
- Content OS Workflow: Four-phase system (Ideate, Plan, Generate, Distribute) with Kanban boards and visual canvas
- Trend Research: Google Trends and TikTok trend analysis for content ideation
- Ad Spy + Decode: Competitor ad analysis from Meta and TikTok ad libraries — proven winners are decoded into reusable Skills (hook, script, scenes, audio, CTA) you can apply to your own product
- Social Publishing: Direct scheduling and publishing to YouTube, TikTok, Instagram, and Twitter/X
- Visual Pipeline Editor: Node-based workflow builder for chaining AI operations
- Model Marketplace: Browse and compare all available AI models with filtering

## Pricing

AI Content Drop uses a credit-based system with monthly subscriptions:
- Free: 10 credits, once, for image generation with Z-Image, FLUX.2 or Seedream 4.5 (video needs a paid plan)
- Starter: $19/month
- Professional: $49/month
- Ultra: $99/month
- Business: $299/month

Checkout runs on Stripe. Seedance 2.5 and the Seedance 2.0 family are billed per second of output; most other models cost a flat amount per generation, charged only on success. Unused plan credits do not roll over; top-up credits carry over. Cancel any time from the account page. The first payment for a plan is refundable within 7 days if none of its credits were used ([refund policy](https://aicontentdrop.com/refund-policy)). Full plan, credit and per-model pricing: [pricing.md](https://aicontentdrop.com/pricing.md).

## How It Differs From Competitors

Unlike single-model tools (Runway, Pika, Luma), AI Content Drop aggregates many AI providers into one platform. Unlike pure prompt-to-video tools, it wraps generation in a full content workflow — from trend research and planning through generation to social publishing and analytics. The Chat-to-Ads Studio interface makes it accessible to marketers who don't want to learn prompt engineering.

Compared to Creatify, Synthesia, or HeyGen, AI Content Drop offers a wider model selection, a credit-based pay-per-use system rather than seat-based pricing, and an integrated content planning workflow.

## Technology

- Frontend: React 18, TypeScript, Vite, Tailwind CSS
- Backend: Express.js, TypeScript, Supabase (PostgreSQL)
- AI Providers: Google (Gemini, Veo, Nano Banana Pro, Imagen), OpenAI (Sora, GPT Image), Anthropic (Claude), Kuaishou (Kling), Alibaba (Qwen), Bytedance (Seedance, Seedream), xAI (Grok Imagine), Black Forest Labs (FLUX), Stability AI
- Payments: Stripe
- Hosting: Railway, Cloudflare CDN

## Company

AI Content Drop (formerly Vidify) is an independent SaaS product. Website: https://aicontentdrop.com

## Links

- [Homepage](https://aicontentdrop.com) — chat-to-video studio and product overview
- [Pricing](https://aicontentdrop.com/pricing) — plans, credits, and what each tier includes
- [About](https://aicontentdrop.com/about) — team, mission, and company background
- [Contact](https://aicontentdrop.com/contact) — support, sales, abuse, and legal channels
- [Plans and prices as JSON](https://aicontentdrop.com/v1/plans) — subscription tiers, credit packs, and the free tier from the same constants checkout bills against. No credential.
- [AI Content Drop Developer API documentation](https://aicontentdrop.com/developers) — REST API, API keys, MCP server
- [Blog](https://aicontentdrop.com/blog) — guides, model comparisons, and tutorials
- [Use cases](https://aicontentdrop.com/use-cases/ecommerce) — ecommerce, agencies, creators, social
- [AI model marketplace](https://aicontentdrop.com/marketplace) — full catalogue with credit pricing
- [Seedance 2.5](https://aicontentdrop.com/seedance-2-5) — ByteDance's Seedance 2.5 online: takes up to 30 seconds, examples, and the credit table for what renders here
- [Seedance 2.0](https://aicontentdrop.com/seedance-2-0) — Seedance 2.0 tiers, examples, and credit costs
- [Kling](https://aicontentdrop.com/models/kling) — Kling 3.0 and earlier Kling versions with credits per clip
- [Veo 3](https://aicontentdrop.com/models/veo-3) — Google Veo 3.1, Veo 3 Fast and Quality with credits per clip
- [SORA](https://aicontentdrop.com/models/sora) — Sora 2 model page
- [Kling vs SORA](https://aicontentdrop.com/compare/kling-vs-sora) and [Veo 3 vs SORA](https://aicontentdrop.com/compare/veo-3-vs-sora) — head-to-head comparisons
- [Best AI video generator](https://aicontentdrop.com/best-ai-video-generator) — ranked model comparison
- [AI UGC video generator](https://aicontentdrop.com/features/ugc) — UGC-style ads with AI avatars and voiceover
- [Free creator tools](https://aicontentdrop.com/tools) — free calculators and generators for captions, hooks, hashtags, scripts, and ad budgets
- Research: [AI video generation statistics](https://aicontentdrop.com/blog/ai-video-generation-statistics-2026) · [AI marketing statistics](https://aicontentdrop.com/blog/ai-marketing-statistics-2026) · [AI content creation statistics](https://aicontentdrop.com/blog/ai-content-creation-statistics-2026) · [Video production cost benchmarks](https://aicontentdrop.com/blog/video-production-cost-benchmarks-2026) — sourced figures with methodology sections

## For Agents

- [Full text corpus](https://aicontentdrop.com/llms-full.txt) — every public article as markdown, one file
- Markdown twin of any public page: append `.md` to the path (e.g. /blog/kling-vs-sora-comparison.md), or send `Accept: text/markdown`
- [auth.md](https://aicontentdrop.com/auth.md) — what an agent may and may not do on a user's behalf
- [Agent skills index](https://aicontentdrop.com/.well-known/agent-skills/index.json) — importable SKILL.md guides (AI video prompting, UGC ad scripts, model selection)
- [Agent Plugin manifest](https://aicontentdrop.com/plugin.json) — conformant [Agent Plugins v1.0.0](https://agent-plugins.org/specification) package. The manifest schema is closed, so the two component types sit where the specification puts them: MCP servers at [/.well-known/agent-plugin-mcp.json](https://aicontentdrop.com/.well-known/agent-plugin-mcp.json) (the package's `mcp.json`), skills under `skills/`. Manifest also at [/.well-known/agent-plugin.json](https://aicontentdrop.com/.well-known/agent-plugin.json).
- [AGENTS.md](https://aicontentdrop.com/AGENTS.md) — how an AI coding agent should work against this API: what is authoritative, what to never assume, and which surface to prefer.
- [SKILL.md](https://aicontentdrop.com/SKILL.md) — the flagship importable skill: pick an AI video model for a brief and price it before generating.
- [Agent plugin skills index](https://aicontentdrop.com/skills/index.json) — fourteen importable [Agent Skills](https://agentskills.io/specification), each at `/skills/<name>/SKILL.md`: `ai-video-model-picker` (choose a model and quote it), `video-prompt-writer` (turn a brief into a prompt that survives the model), `image-generation` (27 image models, draft cheap and finish expensive), `batch-generation-runner` (many generations without melting the rate limit), `api-error-handling` (every error code and whether to retry), `credits-and-plans` (what a job costs and whether the user can afford it), `marketing-studio-ad` (brief, quote, approve, start and poll a finished ad or UGC video through the studio tools), `canvas-board` (plan on a Visual Canvas board: lay out text, media and generation cards, quote and approve each render, hand over the board link), `muse-connector` (drive the MCP server from a host that builds its own client: quote, submit, poll, and surface refusals verbatim), `competitor-ad-research` (search competitor ads, decode a winner into a brief, put it on a board), `video-agent` (hand a whole campaign to Drop, the video agent, and approve each paid step by its exact quote), `recreate-reference` (make a video in the style of a reference clip or image: read it into words, start a drop from that reading, read each piece's score, and approve a retry by its exact quote), `drama-studio` (write, price, render and cut a short vertical drama), `short-drama-writer` (write a short drama as text: hooks, power moves, a shot per beat).
- [A2A agent](https://aicontentdrop.com/a2a) (read-only, no auth) — JSON-RPC 2.0, message/send + tasks/get. Skills: model-cost-lookup, article-research. Card: https://aicontentdrop.com/.well-known/agent-card.json
- [Web Bot Auth directory](https://aicontentdrop.com/.well-known/http-message-signatures-directory) — our outbound crawlers sign their requests; verify them against these keys
- [MCP server](https://aicontentdrop.com/mcp) — Streamable HTTP at `/mcp`; deprecated 2024-11-05 HTTP+SSE at the distinct `/mcp/sse` URL. The catalogue and generation tools are discoverable anonymously. The read tools are callable without a credential: list_models, estimate_credit_cost (returns a signed quote_id), search_articles, get_article. OAuth 2.1 sign-in (protected-resource metadata at https://aicontentdrop.com/.well-known/oauth-protected-resource/mcp) or an API key (`Authorization: Bearer acd_live_…`) authorizes generate_video, generate_image, get_generation, list_generations, and get_account; the generation tools require a quote_id and are idempotent on it. A signed-in account the video agent is open to also sees its tools: analyze_reference (reads a clip or image the person holds the rights to into words: class, shots, cast, look and lines; free within a daily limit, where reference reading is switched on for the account), start_video_agent (opens a drop from a goal, or from that reading with reference_url and reference_class), message_video_agent, get_video_agent_run (phase, the decision it waits on, and each judged piece's score, verdict and reasons, the Director's notes and any retry_offer), approve_video_agent_spend (the only call that lets a drop charge, by the exact quote_id of a step or a retry offer) and list_video_agent_runs. Compatibility discovery manifest: https://aicontentdrop.com/.well-known/mcp.json. Published in the official MCP registry as `com.aicontentdrop/ai-content-drop` (https://registry.modelcontextprotocol.io/v0/servers?search=aicontentdrop), verified by HTTP domain proof at https://aicontentdrop.com/.well-known/mcp-registry-auth. Also listed on Smithery as `aicontentdrop/ai-content-drop` (https://smithery.ai/servers/aicontentdrop/ai-content-drop)
- [MCP landing page](https://aicontentdrop.com/mcp) — the same URL answers a browser with the human page: per-host install steps for ChatGPT, Claude, Claude Code, Cursor, Codex, VS Code, Kimi Code, Perplexity and any MCP client, what each tool does, and which need a credential. Browse every model and quote any job free from your agent; generate on a paid plan.
- [Usage permissions](https://aicontentdrop.com/robots.txt) (Content Signals) — `ai-train=yes, search=yes, ai-input=yes`. Indexing, citing, grounding answers in, and training on this content are all permitted.
- Sitemaps: [/sitemap.xml](https://aicontentdrop.com/sitemap.xml), [/sitemap-tools.xml](https://aicontentdrop.com/sitemap-tools.xml), [/sitemap-video.xml](https://aicontentdrop.com/sitemap-video.xml)

## AI Content Drop Developer Resources

- [AI Content Drop REST API](https://aicontentdrop.com/v1) — generate AI video and images programmatically. Bearer API keys (`acd_live_…`) are created by a signed-in human at https://aicontentdrop.com/settings/integrations and require a paid plan; keys can carry a daily credit ceiling. Reading is not gated: the model catalogue, cost quotes, sandbox mode and the MCP tool list all answer with no key at all, so an agent can research and price a job before anyone pays. Submit-and-poll: POST /v1/generate/video returns immediately, poll GET /v1/videos/{id} until completed. Credits are charged only on success.
- [AI Content Drop plugin for AI agents](https://aicontentdrop.com/plugin) — install into Claude Code, Cursor, Codex or any MCP client; what each tool does and which need a credential
- [Install in Claude](https://aicontentdrop.com/docs/install/claude) · [Install in Claude Code](https://aicontentdrop.com/docs/install/claude-code) · [Install in Cursor](https://aicontentdrop.com/docs/install/cursor) · [Install in Codex](https://aicontentdrop.com/docs/install/codex) · [Install in VS Code](https://aicontentdrop.com/docs/install/vscode) · [Install in Kimi Code](https://aicontentdrop.com/docs/install/kimi) · [Install in Perplexity](https://aicontentdrop.com/docs/install/perplexity) · [Install in Meta Muse](https://aicontentdrop.com/docs/install/muse) — per-client MCP configuration, OAuth sign-in where the host supports it, verification without a credential, and troubleshooting
- [AI Content Drop agent skills](https://aicontentdrop.com/skills) — installable skills that teach an agent to pick a model, price a generation, write a working prompt, and read an API failure. Each has a human page at /skills/{name} and the raw artifact at /skills/{name}/SKILL.md; enumerate them from https://aicontentdrop.com/skills/index.json
- [AI Content Drop developer documentation index](https://aicontentdrop.com/docs) — every reference below, one page each, with a markdown twin at the same path + `.md`
- [AI Content Drop REST API reference](https://aicontentdrop.com/docs/api) — every endpoint, request and response shape
- [AI Content Drop API authentication](https://aicontentdrop.com/docs/authentication) — API keys, agent self-registration (`agent_auth` with `register_uri`, `claim_uri`, `revocation_uri`), WWW-Authenticate discovery. `identity_assertion` / `id-jag` are not supported and [auth.md](https://aicontentdrop.com/auth.md) says why.
- [AI Content Drop MCP server documentation](https://aicontentdrop.com/docs/mcp) — tools, annotations, and the ui:// widgets
- [AI Content Drop GraphQL API](https://aicontentdrop.com/docs/graphql) — Relay pagination, typed errors, async jobs, deprecation policy
- [Sandbox mode](https://aicontentdrop.com/docs/sandbox) — rehearse a generation with no key and no credits
- [Batch operations](https://aicontentdrop.com/docs/batch) — many quotes or reads in one request
- [Rate limits](https://aicontentdrop.com/docs/rate-limits) and [error codes](https://aicontentdrop.com/docs/errors) — every response carries `RateLimit-Limit`, `RateLimit-Remaining`, `RateLimit-Reset` and `RateLimit-Policy`; a 429 adds `Retry-After` (seconds) and `error.retry_after_seconds`. Error codes are lower snake_case in `{ "error": { "code", "message" } }`.
- [Versioning and deprecation policy](https://aicontentdrop.com/docs/versioning) — `/v1` is additive-only, every response carries `API-Version`, retirement is announced with RFC 9745 `Deprecation` + RFC 8594 `Sunset` at least 180 days ahead. Nothing on /v1 is deprecated today.
- [Agent-initiated signup](https://aicontentdrop.com/v1/auth/register) — `POST {"email": "..."}` sends the HUMAN a sign-in link and returns 202 `human_action_required` with a line to relay to them. No password parameter: an agent never holds a user's password. The account holds **0 credits** until the human clicks the link, which confirms the address and releases the 10 free credits. 5/hour per IP.
- [Agent onboarding descriptor](https://aicontentdrop.com/.well-known/agent-onboarding.json) — runnable verification for zero-auth reads, the one-time free tier, sandbox mode, and anonymous `POST /agent/auth/register` read tokens. Minting the `acd_live_…` generation key is explicitly marked as a signed-in human action, not anonymously verifiable. Start here if you are deciding what can proceed without a human.
- [AI Content Drop pricing, credits, and per-model costs](https://aicontentdrop.com/pricing.md) — plans and what every model costs, as plain markdown
- [Product overview page](https://aicontentdrop.com/developers)
- [Official CLI + SDKs](https://github.com/aicontentdrop/aicontentdrop) — the `aicontentdrop` package in three ecosystems, MIT: [npm](https://www.npmjs.com/package/aicontentdrop) (TypeScript SDK + `acd` CLI, `npm install -g aicontentdrop`), [PyPI](https://pypi.org/project/aicontentdrop/) (Python 3.9+, zero runtime dependencies, `pip install aicontentdrop`), and [Go](https://pkg.go.dev/github.com/aicontentdrop/aicontentdrop/go) (standard library only, `go get github.com/aicontentdrop/aicontentdrop/go`). The GitHub repo carries the source for both, AGENTS.md, the skills, and the agent plugin manifests.
- [OpenAPI 3.1 specification](https://aicontentdrop.com/openapi.json) — the REST API surface
- [Content-surface OpenAPI spec](https://aicontentdrop.com/openapi-content.json) — llms.txt, markdown twins, sitemaps
- [GraphQL API](https://aicontentdrop.com/graphql) — POST JSON; introspection is open with no credential. SDL: [/graphql/schema.graphql](https://aicontentdrop.com/graphql/schema.graphql). Relay connections, typed `UserError` payloads, async `GenerationJob` polling via `job(id:)`, `@cost` and `@rateLimit` directives with a 1000-point per-request budget, and a deprecation register at `apiInfo { deprecations }`.
- [GraphQL subscriptions](https://aicontentdrop.com/graphql/stream) — `generationProgress(id:)` over Server-Sent Events (graphql-sse distinct-connections mode): POST /graphql with `Accept: text/event-stream`, or GET /graphql/stream?query=…. No WebSocket endpoint.
- [Batch endpoint](https://aicontentdrop.com/v1/batch) — up to 20 read operations in one POST, no key required
- [NLWeb /ask](https://aicontentdrop.com/ask) — natural-language query endpoint, read-only, no key: POST `{"query":"which video models cost under 15 credits?"}` returns Schema.org-shaped JSON
- [API catalogue (RFC 9727)](https://aicontentdrop.com/.well-known/api-catalog) — every API we publish, in one linkset
- [Agentic Resource Discovery catalogue](https://aicontentdrop.com/.well-known/ai-catalog.json) — MCP servers, APIs, agents, skills
- CLI + SDKs: the `aicontentdrop` package is on npm (TypeScript SDK + `acd` CLI — `npm install -g aicontentdrop`, or keyless reads via `npx aicontentdrop models`), on PyPI (Python, `pip install aicontentdrop`, no runtime dependencies), and as a Go module (`go get github.com/aicontentdrop/aicontentdrop/go`, standard library only). Every call either one wraps is one of the documented HTTP calls below, so the REST endpoints remain a full alternative.
- [Docs-only MCP server](https://aicontentdrop.com/mcp/docs) — a SEPARATE server (`serverInfo.name: aicontentdrop-docs`) with two tools, search_articles and get_article, and nothing that generates. The product server above is `aicontentdrop` and carries every tool; do not compare one against the other's manifest.
- [Machine-readable homepage](https://aicontentdrop.com/?mode=agent) — what we do, pricing, and every agent endpoint as JSON
- Scoped context: [/developers/llms.txt](https://aicontentdrop.com/developers/llms.txt), [/tools/llms.txt](https://aicontentdrop.com/tools/llms.txt), [/blog/llms.txt](https://aicontentdrop.com/blog/llms.txt)
- Zero-auth endpoints (no key at all): `GET /v1` (index), `GET /v1/models` (catalogue with credit costs), `GET /v1/models/{id}/cost`, `POST /v1/models/cost` (batch quote), `POST /ask`, `POST /v1/batch` (read batch), `POST /graphql` (including introspection), `POST /ask`, `POST /agent/auth/register`, and `POST /v1/generate/*` with `X-Sandbox: true`.
- Idempotency: send `Idempotency-Key` on generation POSTs so a retry cannot double-charge.
- Sandbox: send `X-Sandbox: true` to `POST /v1/generate/video` or `/v1/generate/image` to rehearse generation without spending credits. It needs NO API key, so the response shape is observable before anyone mints a credential.
- Async: generation returns `202` with a `Location` header to poll; list endpoints page by `cursor`.
- Rate limits: 120 requests/minute per IP on the standard /v1 tier, or 600 with a valid `acd_agent_` read token; standard RateLimit-* response headers disclose the selected tier and a 429 carries Retry-After (seconds).
- Versioning: the version is in the URL path. A breaking change ships as a new path segment; the old one keeps working at least 180 days and is announced with `Deprecation` and `Sunset` headers first. Policy: https://aicontentdrop.com/developers#versioning
- Errors are structured JSON: `{"error": {"code", "message"}}` on /v1 reads.
- Account signup itself is still a human flow — see /auth.md before attempting programmatic registration.