JigsawStack: Serving up a variety of small, specialized AI model APIs

General Introduction

JigsawStack is an AI tooling platform for developers. It provides a variety of API services for small AI models to help users quickly integrate features into their applications. The core features are fast speed, global latency as low as 200 ms, and availability as high as 99.51 TP3T. The platform supports useful features such as web crawling, speech-to-text, image recognition, etc. Pricing is per-use, flexible and cost-controlled. Officially, JigsawStack received $500,000 in seed funding in October 2024, backed by Antler. It is suitable for tech teams and developers who need an efficient AI solution.

Function List

• Web page scraping (AI Scrape): Extract web page data, such as news headlines or product prices, by prompting and return structured results.
• Speech-to-Text: Convert audio files to text, optimized for speed based on the Whisper 3 model.
• Image Recognition (VOCR): Extract text or specific information from an image, such as the amount on an invoice.
• Text-to-SQL (Text-to-SQL):: Simplify database operations by converting natural language into SQL query statements.
• Real-time search (AI Search):: Search the web for content, returning accurately sorted results and an overview of AI.
• Text-to-Speech:: Convert text to natural speech, supporting multiple languages.
• Translation services (Translate):: Support for translation into more than 100 languages, taking into account contextual accuracy.
• Predictive modeling (AI Prediction): One-time high-precision prediction of time series data without pre-training.
• Validation AI: Check spam, email validity, sentiment analysis, etc. Support localized data validation.
• Prompt Engine:: Automatically optimize and run the best large language models to support large-scale usage.

Using Help

JigsawStack is designed with the goal of allowing developers to quickly implement features through a simple API. Below is a detailed guide to using it.

Installation process

JigsawStack provides REST API and SDK to support multiple programming languages. The installation steps are as follows:

1. Register and get API key
   Go to https://jigsawstack.com and click "Start for free" to register. After logging in, generate a key on the "API Keys" page of https://jigsawstack.com/dashboard, e.g.: sk39wo393.... .32ncsmw9339RNj3.
2. Installing the SDK
   • JavaScript user to run:

     npm i jigsawstack
     

   • Python user to run:

     pip install jigsawstack
     

3. Initializing the SDK
   Take JavaScript for example:

   import { JigsawStack } from "jigsawstack" ;
   const jigsaw = new JigsawStack({ apiKey: "Your key" });

Main function operation flow

1. Web page scraping (AI Scrape)

This function can quickly extract web page data. Operation steps:

• Prepare the web page URL, e.g. https://news.ycombinator.com/show.
• Call Interface:

  const result = await jigsaw.web.ai_scrape({
  url: "https://news.ycombinator.com/show", element_prompts: ["post title", "post points"]]
  element_prompts: ["post title", "post points"]
  });
  console.log(result);
  

• Returns a sample result:

  {
  
  "post points": ["150"]
  }
  

• draw attention to sth.: element_prompts Specify the content to be extracted, which is more efficient and concise.

2. Speech-to-Text

Based on the Whisper 3 model, it can transcribe audio quickly. Operation steps:

• Prepare the audio URL, e.g.  The
• Call Interface:

  const text = await jigsaw.audio.speech_to_text({
  url: "https://example.com/meeting.mp3"
  }); console.log(text); console.log(text); console.log(text)
  console.log(text);
  

• Example return result: "New feature development was discussed in today's meeting."
• take note of: Supports formats such as MP3, WAV, etc., with files not exceeding the size limit.

3. Image recognition (VOCR)

Extract structured data from images. Operation Steps:

• Prepare the image URL, e.g. https://jigsawstack.com/preview/vocr-example.jpg.
• Call Interface:

  const data = await jigsaw.vision.vocr({
  url: "https://jigsawstack.com/preview/vocr-example.jpg",
  prompt: ["total_price", "tax"]
  });
  console.log(data);
  

• Returns a sample result:

  {
  "total_price": "$50.00",
  "tax": "$5.00"
  }
  

4. Real-time search (AI Search)

Search the web and return accurate results. Operation Steps:

• Call Interface:

  const search = await jigsaw.web.search({
  query: "Best Restaurants in San Francisco"
  }); console.log(search);
  console.log(search);
  

• Returns structured results and AI-generated overviews.

5. Translation services (Translate)

Supports multi-language translation, taking context into account. Operational steps:

• Call Interface:

  const translated = await jigsaw.translate({
  text: ["Hey, how are you?"], target_language: "zh
  target_language: "zh"
  });
  console.log(translated);
  

• Return result: ["Hey, how are you?"]

Management and Commissioning

• instrument panels: View real-time logs and analyze API usage at https://jigsawstack.com/dashboard, including IP, location, and more than 30 other data points.
• key security: Supports domain whitelisting and access permission settings to restrict the key to call only specific APIs.
• (computer) file: Visit https://jigsawstack.com/docs for a detailed API reference and sample code.
• Team Management: Supports multiplayer collaboration, inviting members and assigning roles.

Tips for use

• The free version is suitable for testing, with a monthly limit, and upgrading is recommended for production environments.
• API responses are consistent and all services use a uniform request and return format.
• Global deployment of 99+ edge nodes ensures low latency and high availability.

application scenario

1. News data collection
   Use the web crawler function to extract titles and ratings and analyze trending topics.
2. Voice Conference Organizer
   Use the speech-to-text feature to convert meeting recordings into text for archiving.
3. financial automation
   Use the image recognition function to extract data from invoices and import them into the system.
4. Multi-language support
   Add real-time translation to apps with translation services.

QA

1. What languages does JigsawStack support?
   The API is compatible with JavaScript, Python, PHP and other languages, and translations support more than 100 languages.
2. What are the limitations of the free version?
   The number of calls is limited, see https://jigsawstack.com/pricing.
3. How is data security ensured?
   Data transmission is encrypted and logs do not store sensitive content, in compliance with privacy policies.
4. Can it be integrated into an existing framework?
   Yes, support for LangChain, Zapier, etc. and Vercel and Supabase integration coming soon.