Kanto Labs

Kanto Labs

Small, reliable tools for video, images and AI pipelines.

Pull keyframes and thumbnails from video files, turn a video into context an AI model can read, get a shot list with cut timecodes, pull the slides out of a lecture, read the text shown on screen, break down an ad video into hook, shots and CTA, write captions and alt text for images, upscale images 2x or 4x. Each tool does one job, runs on the Apify platform, and charges only for results it actually delivers.

How a Kanto Labs tool bills Four inputs go into a tool. Three come out as result rows charged a fraction of a cent each. One unreachable input comes out as an error row that costs nothing. INPUT trailer.mp4 webinar.mp4 product.jpg dead-link.mp4 RESULT {8 frames}$.016 {3 scenes}$.014 {4x image}$.010 {error}$0 RUN TOTAL $0.040 FAILED INPUTS free

How you pay

Pay per result. Failures are free.

Every tool uses Apify's pay-per-event billing: you are charged a fixed, published price for each result it delivers, and nothing for inputs it could not process.

A price per resultPer frame, per keyframe, per video minute, per image. Fractions of a cent, published on each tool's page. No subscription, no seats.
Failures cost $0Refused links, broken downloads and unreadable files come back as rows with an error message, so you can see what happened, and they are not charged.
You set the ceilingSet a maximum cost per run and the tool stops cleanly when it is reached. A long list can never cost more than you decided.
Price per result for each tool
ToolPriceCharged per
Video Frame Extractor$0.002frame extracted
Video to LLM Context$0.002keyframe extracted; $0.008 per audio minute
Video to Slides$0.004slide; $0.001 per video minute
Video OCR$0.002frame with text
Video Ad Creative Breakdown$0.025video (up to 3 minutes); $0.01 per extra minute
Video Scene Detection$0.018video minute scanned; $0.002 per clip (optional)
AI Image Upscaler$0.01image upscaled; $0.06 per large image
AI Image Caption & Alt Text Generator$0.004image captioned

Use from AI assistants (MCP)

Ask your assistant. It calls the tool.

The Kanto Labs MCP server plugs our core video and image tools into Claude, Cursor, VS Code Copilot, Windsurf and any other MCP client. Ask “Summarize this webinar recording scene by scene” or “Upscale this product photo 4x” and the assistant runs the right tool on your own Apify account.

  • Every call has a spending cap (default $1.00), so an assistant cannot run up a bill by accident
  • Each result states what the call cost
  • Runs on your machine and talks only to the Apify API
Claude Code
claude mcp add kanto-labs -e APIFY_TOKEN=apify_api_your_token_here -- npx -y github:kanto-labs/kanto-labs-mcp
claude_desktop_config.json
{
  "mcpServers": {
    "kanto-labs": {
      "command": "npx",
      "args": ["-y", "github:kanto-labs/kanto-labs-mcp"],
      "env": { "APIFY_TOKEN": "apify_api_your_token_here" }
    }
  }
}

Bigger projects

Need more than one workflow?

Fixed-scope projects such as lead intake to CRM, document processing or internal tools start from $1,000. Tell us what you need and we reply with a written scope and price.

Blog

Tutorials

Practical write-ups on web data, documents and LLM pipelines. Latest: How to convert PDF to Markdown for LLMs in Python.