<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Erik Milošević on Medium]]></title>
        <description><![CDATA[Stories by Erik Milošević on Medium]]></description>
        <link>https://medium.com/@Erik_Milosevic?source=rss-f4e8d1747a77------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*I8SsceTWEdZkkr1FRq-aUA.png</url>
            <title>Stories by Erik Milošević on Medium</title>
            <link>https://medium.com/@Erik_Milosevic?source=rss-f4e8d1747a77------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 12:41:57 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@Erik_Milosevic/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[How to Use Milvus as a Vector Database in LangChain Guide]]></title>
            <link>https://medium.com/@Erik_Milosevic/how-to-use-milvus-as-a-vector-database-in-langchain-guide-9bf001edf03b?source=rss-f4e8d1747a77------2</link>
            <guid isPermaLink="false">https://medium.com/p/9bf001edf03b</guid>
            <category><![CDATA[langchain]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[large-language-models]]></category>
            <category><![CDATA[programming]]></category>
            <dc:creator><![CDATA[Erik Milošević]]></dc:creator>
            <pubDate>Wed, 20 Aug 2025 09:26:26 GMT</pubDate>
            <atom:updated>2025-08-20T09:26:26.227Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*3UJLbkwdmsBQ6QxT" /></figure><p>Let’s talk about something that we all face during development: API Testing with Postman for your Development Team.</p><p>Yeah, I’ve heard of it as well, <a href="https://www.reddit.com/r/webdev/comments/kl45qq/what_the_hell_happened_to_postman/">Postman is getting worse</a> year by year, but, you are working as a team and you need some collaboration tools for your development process, right? So you paid Postman Enterprise for…. $49/month.</p><p>Now I am telling you: You Don’t Have to:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*uIjDlxDTBB3HRqHZ.png" /></figure><p>That’s right, APIDog gives you all the features that comes with Postman paid version, at a fraction of the cost. Migration has been so easily that you only need to click a few buttons, and APIDog will do everything for you.</p><p><a href="https://bit.ly/3TdVn6U?source=post_page-----f06877ddde99---------------------------------------">How to Migrate Postman Collections/Environments to Apidog</a></p><p><strong>APIDog</strong> has a comprehensive, easy to use GUI that makes you spend no time to get started working (If you have migrated from Postman). It’s elegant, collaborate, easy to use, with Dark Mode too!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*wbRgyd1cjHs-WjnO.png" /></figure><p>Want a Good Alternative to Postman? APIDog is definitely worth a shot. But if you are the Tech Lead of a Dev Team that really want to dump Postman for something Better, and Cheaper, Check out APIDog!</p><p><a href="https://bit.ly/4ebMDGm?source=post_page-----f06877ddde99---------------------------------------">Apidog: All-in-One API Platform: Design, Debug, Mock, Test, and Document</a></p><h3>Understanding the Basics of Vector Databases</h3><p>A vector database is designed to handle large-scale high-dimensional vector data, which is essential for applications like machine learning, natural language processing, and image recognition. Milvus is a prominent example of such a database, providing advanced functionalities for efficient similarity search and embedding management. Langchain, a library specifically designed for building applications powered by language models, can harness the capabilities of Milvus to manage these vectors effectively.</p><h3>Overview of Milvus</h3><p>Milvus is an open-source vector database capable of storing, indexing, and searching high-dimensional vectors generated by deep learning models. Its architecture is optimized for faster retrieval and scalability, making it well-suited for applications that require quick access to similar data embeddings. Using Milvus involves defining an environment where you can efficiently create, manage, and query vector data, facilitating the development of intelligent applications.</p><h3>Installing Milvus</h3><p>To begin using Milvus in your Langchain project, you must first install and set up Milvus. You can do this either on your local machine using Docker or on a cloud service platform. Here, we will provide step-by-step instructions for both approaches.</p><h4>Option 1: Docker Installation</h4><ol><li><strong>Install Docker</strong>: Ensure that you have Docker installed on your system. You can download it from the <a href="https://www.docker.com/products/docker-desktop">official Docker website</a>.</li><li><strong>Pull the Milvus Docker image</strong>: Open your terminal and run the following command:</li></ol><ul><li>docker pull milvusdb/milvus:v2.0.0</li></ul><ol><li><strong>Run Milvus</strong>: After pulling the image, you can run Milvus using the following command:</li></ol><ul><li>docker run -d --name milvus_cpu_2.0.0 \ -p 19530:19530 \ -p 19121:19121 \ milvusdb/milvus:v2.0.0</li></ul><ol><li><strong>Check if Milvus is running</strong>: After starting the container, ensure that your Milvus server is running by accessing <a href="http://localhost:19530.">http://localhost:19530.</a></li></ol><h4>Option 2: Cloud Installation</h4><p>For cloud installation, services such as AWS or GCP provide options for running Milvus in managed environments. Each cloud service may have different steps, but generally, you would:</p><ol><li>Select a compute instance.</li><li>Set up the networking for port access.</li><li>Launch Milvus following the relevant cloud provider’s instructions (often available in their documentation).</li></ol><h3>Integrating Langchain with Milvus</h3><p>With Milvus installed and operational, the next step involves integrating it with Langchain. The Langchain library allows developers to create applications that require an understanding of human language while leveraging the capabilities of various external databases, including Milvus.</p><h4>Step 1: Install Langchain</h4><p>You must install the Langchain library, which can be done using pip. In your terminal or command line, run:</p><pre>pip install langchain</pre><h4>Step 2: Setting Up Connection to Milvus</h4><p>Before making any queries, you must establish a connection between Langchain and Milvus. The following code snippet demonstrates how to do this:</p><pre>from langchain.vectorstores import Milvus<br>from pymilvus import Collection, connections<br><br># Connect to the Milvus server<br>connections.connect(&quot;default&quot;, host=&quot;localhost&quot;, port=&quot;19530&quot;)<br># Initialize your collection (assuming a collection named &#39;example_collection&#39; already exists)<br>collection = Collection(&quot;example_collection&quot;)<br>vector_store = Milvus(collection=collection)</pre><p>In this setup, you are connecting to your Milvus instance and creating a reference to a specific collection where your vectors will be stored.</p><h3>Creating and Inserting Vectors</h3><p>To leverage Milvus effectively, you need to prepare your high-dimensional vectors, typically obtained from embedding models such as BERT or OpenAI’s GPT. Once you have these vectors, the next step is to explore how to create and insert them into Milvus.</p><h4>Step 1: Prepare Your Data</h4><p>For demonstration purposes, let’s assume you have some sentence data you want to convert into vectors. You can use an embedding model as follows:</p><pre>import numpy as np<br>from sentence_transformers import SentenceTransformer<br><br># Load the pre-trained embedding model<br>model = SentenceTransformer(&#39;paraphrase-MiniLM-L6-v2&#39;)<br># Sample data<br>documents = [&quot;This is a dog.&quot;, &quot;This is a cat.&quot;, &quot;Dogs are great pets.&quot;, &quot;Cats are independent animals.&quot;]<br># Generate embeddings<br>embeddings = model.encode(documents).tolist()</pre><p>Here, the model.encode function converts each sentence into a high-dimensional vector.</p><h3>Step 2: Insert Vectors into Milvus</h3><p>Once the vectors are ready, use the following code to insert them into your Milvus collection:</p><pre># Prepare ids<br>ids = [i for i in range(len(embeddings))]<br><br><br># Inserting vectors<br>insert_result = vector_store.add(embeddings, ids)<br>print(&quot;Insert Result: &quot;, insert_result)</pre><p>This inserts the vectors into the Milvus collection while associating them with unique IDs. In most scenarios, these IDs would correspond to the source documents for easy data retrieval.</p><h3>Querying Vectors in Milvus</h3><p>After successfully inserting vectors, you can perform similarity searches to find relevant data based on input queries. Milvus allows for efficient nearest neighbor searches through its vector indexing capabilities.</p><h4>Step 1: Performing Vector Searches</h4><p>To conduct a search in Milvus, you need to convert your query into a vector format as demonstrated in the embedding preparation step. Here’s how you can execute a similarity search:</p><pre># Prepare a query vector<br>query = &quot;I have a pet dog.&quot;<br>query_vector = model.encode(query).tolist()<br><br># Perform search<br>k = 2  # Number of nearest neighbors to retrieve<br>search_results = vector_store.similarity_search(query_vector, k=k)<br># Display results<br>for result in search_results:<br>    print(&quot;Document: &quot;, documents[result.id], &quot; Score: &quot;, result.score)</pre><p>This code snippet conducts a search for the top k most similar vectors to the query vector. The results will show both the matched document and its corresponding similarity score.</p><h3>Advanced Operations: Vector Management</h3><p>While basic insert and query functions serve common use cases, Milvus supports advanced operations like updating and deleting vectors, as well as maintaining index configurations to enhance search performance.</p><h4>Updating Vectors</h4><p>To update an existing vector, you can retrieve the vector using its ID and replace it with a new vector:</p><pre># New vector for a specific document (ID = 0)<br>new_vector = model.encode(&quot;I adopted a puppy.&quot;).tolist()<br><br># Update the vector in Milvus<br>vector_store.update(0, new_vector)</pre><p>Make sure that the structure of the new vector matches what is expected within the collection.</p><h4>Deleting Vectors</h4><p>Deleting vectors can be crucial for data management. You can remove vectors based on their IDs:</p><pre># Delete a vector by ID<br>vector_store.delete(ids=[0])</pre><p>This simplicity in management allows for dynamic updates within your datasets, ensuring that your vector database remains relevant and useful.</p><h4>Scaling Milvus in Production</h4><p>When considering Milvus for production use, several factors come into play, namely performance optimization and scaling. Milvus supports multiple indexing techniques — like IVF, HNSW, and ANNOY — each suitable for different types of queries and workloads.</p><h3>Steps to Optimize Performance</h3><ol><li><strong>Select the Right Index Type</strong>: Understand the nature of your data and choose an indexing algorithm that balances between speed and accuracy for your specific application.</li><li><strong>Configure Index Parameters</strong>: Customize index parameters based on the indexing strategy you chose. For instance, IVF can be modified for the number of centroids to balance between search speed and result accuracy.</li><li><strong>Batch Insertions</strong>: For large datasets, consider batch insert operations rather than inserting vectors one at a time.</li><li><strong>Horizontal Scaling</strong>: If needed, deploy Milvus in a distributed environment, allowing for multiple nodes to handle increasing loads.</li></ol><p>By implementing these strategies, you can ensure that your Milvus instance has the scalability and performance required to meet production demands.</p><h3>Conclusion</h3><p>In this comprehensive exploration of how to use Milvus as a vector database within the Langchain framework, we have covered foundational concepts, installation processes, data management, and querying methodologies. By leveraging the powerful integration between Langchain and Milvus, developers can create intelligent applications capable of processing and understanding complex datasets efficiently. As the landscape of AI continues to evolve, integrating robust solutions such as Milvus will remain essential for creating performant and intelligent systems.</p><p>Let’s talk about something that we all face during development: API Testing with Postman for your Development Team.</p><p>Yeah, I’ve heard of it as well, <a href="https://www.reddit.com/r/webdev/comments/kl45qq/what_the_hell_happened_to_postman/">Postman is getting worse</a> year by year, but, you are working as a team and you need some collaboration tools for your development process, right? So you paid Postman Enterprise for…. $49/month.</p><p>Now I am telling you: You Don’t Have to:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*FLeI-tv4fub6nplF.png" /></figure><p>That’s right, APIDog gives you all the features that comes with Postman paid version, at a fraction of the cost. Migration has been so easily that you only need to click a few buttons, and APIDog will do everything for you.</p><p><a href="https://bit.ly/3TdVn6U?source=post_page-----f06877ddde99---------------------------------------">How to Migrate Postman Collections/Environments to Apidog</a></p><p><strong>APIDog</strong> has a comprehensive, easy to use GUI that makes you spend no time to get started working (If you have migrated from Postman). It’s elegant, collaborate, easy to use, with Dark Mode too!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*tnDvBpSOYwV-dbIs.png" /></figure><p>Want a Good Alternative to Postman? APIDog is definitely worth a shot. But if you are the Tech Lead of a Dev Team that really want to dump Postman for something Better, and Cheaper, Check out APIDog!</p><p><a href="https://bit.ly/4ebMDGm?source=post_page-----f06877ddde99---------------------------------------">Apidog: All-in-One API Platform: Design, Debug, Mock, Test, and Document</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9bf001edf03b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Use JavaScript with Langchain: A Step-by-Step Guide]]></title>
            <link>https://medium.com/@Erik_Milosevic/how-to-use-javascript-with-langchain-a-step-by-step-guide-3267297a06b6?source=rss-f4e8d1747a77------2</link>
            <guid isPermaLink="false">https://medium.com/p/3267297a06b6</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[large-language-models]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[langchain]]></category>
            <dc:creator><![CDATA[Erik Milošević]]></dc:creator>
            <pubDate>Wed, 20 Aug 2025 09:18:45 GMT</pubDate>
            <atom:updated>2025-08-20T09:18:45.112Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*GPt459rTQD14h34k" /></figure><p>Let’s talk about something that we all face during development: API Testing with Postman for your Development Team.</p><p>Yeah, I’ve heard of it as well, <a href="https://www.reddit.com/r/webdev/comments/kl45qq/what_the_hell_happened_to_postman/">Postman is getting worse</a> year by year, but, you are working as a team and you need some collaboration tools for your development process, right? So you paid Postman Enterprise for…. $49/month.</p><p>Now I am telling you: You Don’t Have to:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*_bZXk2wh9OdAAvvm.png" /></figure><p>That’s right, APIDog gives you all the features that comes with Postman paid version, at a fraction of the cost. Migration has been so easily that you only need to click a few buttons, and APIDog will do everything for you.</p><p><a href="https://bit.ly/3TdVn6U?source=post_page-----35c6de39f997---------------------------------------">How to Migrate Postman Collections/Environments to Apidog</a></p><p><strong>APIDog</strong> has a comprehensive, easy to use GUI that makes you spend no time to get started working (If you have migrated from Postman). It’s elegant, collaborate, easy to use, with Dark Mode too!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ir_0kb5Qk9bQ41-B.png" /></figure><p>Want a Good Alternative to Postman? APIDog is definitely worth a shot. But if you are the Tech Lead of a Dev Team that really want to dump Postman for something Better, and Cheaper, Check out APIDog!</p><p><a href="https://bit.ly/4ebMDGm?source=post_page-----35c6de39f997---------------------------------------">Apidog: All-in-One API Platform: Design, Debug, Mock, Test, and Document</a></p><h3>How to Use JavaScript with LangChain</h3><p>LangChain is a robust framework that allows developers to create applications utilizing large language models (LLMs) in various ways, including but not limited to chatbots and text processing pipelines. Its versatility, especially when combined with JavaScript, opens up a plethora of possibilities for developers. In this article, we will explore how to effectively use JavaScript with LangChain, covering installation, basic constructs, integrations with various LLMs, and real-world application examples.</p><h4>Setting Up Your Environment</h4><p>To start using JavaScript with LangChain, you must ensure that you have an appropriate development environment set up. This requires Node.js and npm (Node Package Manager).</p><h4>Step 1: Install Node.js</h4><ol><li><strong>Download Node.js</strong>: Visit <a href="https://nodejs.org/">the Node.js website</a> and download the latest LTS version suited for your operating system.</li><li><strong>Install Node.js</strong>: Follow the installation instructions provided on the website. Upon successful installation, you can verify it by opening your terminal and running: node -v</li></ol><p>3. This command returns the installed Node.js version.</p><h4>Step 2: Initialize a New Project</h4><p><strong>1. Create a new directory</strong>: In your terminal, run:</p><ul><li>mkdir langchain-js-app cd langchain-js-app</li></ul><p><strong>2. Initialize npm</strong>: Execute the command:</p><ul><li>npm init -y</li></ul><p><strong>3. </strong>This creates a package.json file, which will manage the dependencies for your project.</p><h4>Step 3: Install LangChain</h4><ol><li><strong>Install LangChain</strong>: Now, you can install LangChain by running:</li></ol><ul><li>npm install langchain</li></ul><p>This will add LangChain and its dependencies to your project, enabling you to start creating applications that utilize LLMs.</p><h3>Basic Constructs in LangChain</h3><p>LangChain provides various classes and structures that allow you to implement language model capabilities. Some fundamental components include LLM, PromptTemplate, and Chain.</p><h3>Using LLMs</h3><p>To use an LLM, you generally need an API key from the provider you want to work with. Popular options include OpenAI and Hugging Face.</p><h4>Step 1: Get API Access</h4><ol><li><strong>OpenAI</strong>: Sign up at <a href="https://platform.openai.com/signup">OpenAI’s platform</a> to retrieve an API key.</li><li><strong>Hugging Face</strong>: Create an account at <a href="https://huggingface.co/">Hugging Face</a> and get an API key from your settings.</li></ol><h4>Step 2: Integrating LLM in JavaScript</h4><p>Here is a basic example of how you can set up and utilize a language model within LangChain.</p><pre>const { OpenAI } = require(&quot;langchain/llms/openai&quot;);<br><br>const openai = new OpenAI({<br>  apiKey: process.env.OPENAI_API_KEY, // make sure to set your API Key<br>});<br>async function generateText(prompt) {<br>  const response = await openai.call(prompt);<br>  console.log(response);<br>}<br>// Example usage<br>generateText(&quot;What are the benefits of using large language models in software development?&quot;);</pre><p>In this snippet:</p><ul><li>We import the OpenAI class from LangChain.</li><li>The openai instance is created with your API Key referenced from environment variables.</li><li>The generateText function can maintain any prompt you provide, and it will respond accordingly.</li></ul><h3>Creating and Using Prompt Templates</h3><p>Creating prompt templates aids in structuring the input provided to language models effectively.</p><h4>Step 1: Setting Up a PromptTemplate</h4><p>LangChain allows you to create reusable prompt templates. Here’s how you can do it:</p><pre>const { PromptTemplate } = require(&quot;langchain/prompts&quot;);<br><br>const soupPrompt = new PromptTemplate({<br>  template: &quot;What ingredients do you need to make {dish}?&quot;,<br>  inputVariables: [&quot;dish&quot;],<br>});<br>async function generateSoupIngredients() {<br>  const prompt = soupPrompt.format({ dish: &quot;tomato soup&quot; });<br>  await generateText(prompt);<br>}<br>generateSoupIngredients(); // This will ask for ingredients of tomato soup</pre><p>In this code:</p><ul><li>The PromptTemplate class is imported.</li><li>A new prompt template is defined that takes an input variable dish.</li><li>The generateSoupIngredients function formats the prompt and calls the generateText function from the previous example.</li></ul><h3>Building Chains for Complex Workflows</h3><p>LangChain supports the creation of chains, which allow you to combine multiple components in a single workflow.</p><h4>Step 1: Creating a Simple Chain</h4><p>A chain can be as simple or complex as your use case demands. Here’s how you can implement a basic chain with multiple prompts.</p><pre>const { LLMChain } = require(&quot;langchain/chains&quot;);<br>const { OpenAI } = require(&quot;langchain/llms/openai&quot;);<br><br>const openai = new OpenAI({<br>  apiKey: process.env.OPENAI_API_KEY,<br>});<br>// Define prompt templates<br>const firstPrompt = new PromptTemplate({<br>  template: &quot;Summarize the following text: {text}&quot;,<br>  inputVariables: [&quot;text&quot;],<br>});<br>const summarizationChain = new LLMChain({<br>  llm: openai,<br>  prompt: firstPrompt,<br>});<br>// Function to summarize text<br>async function summarizeText(inputText) {<br>  const summary = await summarizationChain.call({ text: inputText });<br>  console.log(summary);<br>}<br>// Example usage<br>summarizeText(&quot;LangChain is a powerful tool for developing applications with LLMs.&quot;);</pre><p>In this example:</p><ul><li>We define a firstPrompt to summarize texts and create a summarizationChain.</li><li>The summarizeText function runs the chain and logs the summary to the console, showcasing the ability to create a seamless workflow.</li></ul><h3>Integrating Additional Tools</h3><p>LangChain is not limited to just LLMs; it also supports integrating tools like memory, document loaders, and vector stores.</p><h4>Step 1: Using Memory in LangChain</h4><p>Memory allows chains to remember past interactions. Here’s a small example to illustrate the concept:</p><pre>const { ConversationChain } = require(&quot;langchain/chains&quot;);<br><br>// Define a conversation chain with memory<br>const conversation = new ConversationChain({<br>  llm: openai,<br>});<br>// Function for handling conversations<br>async function chatWithLLM(input) {<br>  const response = await conversation.call({ input: input });<br>  console.log(response);<br>}<br>// Example chat<br>chatWithLLM(&quot;Tell me about the weather today.&quot;);</pre><p>This code illustrates how to create a conversational agent that retains context through interactions.</p><h4>Step 2: Using Document Loaders</h4><p>If you need to work with datasets or documents, LangChain provides tools for loading and processing documents. Below is a brief example incorporating a document loader:</p><pre>const { DirectoryLoader } = require(&quot;langchain/document_loaders&quot;);<br><br>const loader = new DirectoryLoader(&quot;path/to/documents&quot;);<br>async function loadAndQueryDocuments() {<br>  const documents = await loader.load();<br>  console.log(documents); // Outputs the loaded documents<br>  // Here, you would typically process or manipulate the documents further<br>}<br>// Example usage<br>loadAndQueryDocuments();</pre><p>This example shows how to load documents from a specified directory, making it easier to work with larger datasets.</p><h3>Real-World Applications of LangChain</h3><p>LangChain is versatile and can be used for numerous applications, including but not limited to chatbots, document summarization, and content generation.</p><h4>Example 1: Building a FAQ Bot</h4><p>A common use case for LangChain is creating a FAQ bot that can answer specific queries based on a defined set of documents.</p><pre>const { OpenAI } = require(&quot;langchain/llms/openai&quot;);<br>const { RetrievalQA } = require(&quot;langchain/chains&quot;);<br>const { DirectoryLoader } = require(&quot;langchain/document_loaders&quot;);<br><br><br>const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });<br>const loader = new DirectoryLoader(&quot;path/to/faq/documents&quot;);<br>async function createFAQBot() {<br>  const documents = await loader.load();<br>  const qaChain = new RetrievalQA({ llm: openai, documents });<br>  const answer = await qaChain.call({ query: &quot;What is the return policy?&quot; });<br>  console.log(answer); // The FAQ bot will respond based on the documents<br>}<br>createFAQBot();</pre><h4>Example 2: Text Generation for Marketing</h4><p>You could use LangChain to dynamically generate marketing copy based on user input.</p><pre>const marketingPrompt = new PromptTemplate({<br>  template: &quot;Create a marketing slogan for {product}.&quot;,<br>  inputVariables: [&quot;product&quot;],<br>});<br><br>async function generateMarketingSlogan() {<br>  const prompt = marketingPrompt.format({ product: &quot;organic shampoo&quot; });<br>  const slogan = await openai.call(prompt);<br>  console.log(slogan);<br>}<br>generateMarketingSlogan(); // Outputs a catchy slogan for organic shampoo</pre><p>Both examples demonstrate how LangChain can be used to facilitate specific business needs through natural language generation and processing.</p><p>By utilizing the various capabilities of LangChain within JavaScript, developers can build powerful applications that leverage the strengths of large language models. Following the outlined steps allows you to create a wide array of functionalities, from basic text generation to complex conversational AI, while effectively managing resources, prompts, and documents.</p><p>Let’s talk about something that we all face during development: API Testing with Postman for your Development Team.</p><p>Yeah, I’ve heard of it as well, <a href="https://www.reddit.com/r/webdev/comments/kl45qq/what_the_hell_happened_to_postman/">Postman is getting worse</a> year by year, but, you are working as a team and you need some collaboration tools for your development process, right? So you paid Postman Enterprise for…. $49/month.</p><p>Now I am telling you: You Don’t Have to:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*cjWCJNqrPOl3w9uD.png" /></figure><p>That’s right, APIDog gives you all the features that comes with Postman paid version, at a fraction of the cost. Migration has been so easily that you only need to click a few buttons, and APIDog will do everything for you.</p><p><a href="https://bit.ly/3TdVn6U?source=post_page-----35c6de39f997---------------------------------------">How to Migrate Postman Collections/Environments to Apidog</a></p><p><strong>APIDog</strong> has a comprehensive, easy to use GUI that makes you spend no time to get started working (If you have migrated from Postman). It’s elegant, collaborate, easy to use, with Dark Mode too!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Vr0jurMpZcn0qSrh.png" /></figure><p>Want a Good Alternative to Postman? APIDog is definitely worth a shot. But if you are the Tech Lead of a Dev Team that really want to dump Postman for something Better, and Cheaper, Check out APIDog!</p><p><a href="https://bit.ly/4ebMDGm?source=post_page-----35c6de39f997---------------------------------------">Apidog: All-in-One API Platform: Design, Debug, Mock, Test, and Document</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3267297a06b6" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Run Kimi K2 Inside Claude Code: The Ultimate Open-Source AI Coding Combo]]></title>
            <link>https://medium.com/@Erik_Milosevic/how-to-run-kimi-k2-inside-claude-code-the-ultimate-open-source-ai-coding-combo-22b743b69e5a?source=rss-f4e8d1747a77------2</link>
            <guid isPermaLink="false">https://medium.com/p/22b743b69e5a</guid>
            <category><![CDATA[claude-code]]></category>
            <category><![CDATA[mcp-server]]></category>
            <category><![CDATA[open-source-ai-coding]]></category>
            <category><![CDATA[kimi-k2]]></category>
            <dc:creator><![CDATA[Erik Milošević]]></dc:creator>
            <pubDate>Wed, 20 Aug 2025 09:08:16 GMT</pubDate>
            <atom:updated>2025-08-20T09:08:16.968Z</atom:updated>
            <content:encoded><![CDATA[<p>The AI coding landscape just got a major upgrade. Developers are now running Moonshot AI’s Kimi K2 model directly within the Claude Code environment — combining the raw power of a trillion-parameter open-source LLM with the developer-centric polish of Claude’s interface. The result? A next-level, agentic coding experience that’s both accessible and insanely effective.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*dpdrBrIaFAYcv39E.png" /></figure><p>This guide will show you why this combo is so powerful, what makes Kimi K2 a standout in open-source AI, and how you can set it up in minutes. Let’s dive in.</p><blockquote><strong><em>Pro Tip: </em></strong><em>Want to supercharge your AI coding workflow? </em><a href="https://bit.ly/41ItNSS"><em>Apidog MCP Server</em></a><em> lets you connect your API specs directly to AI-powered IDEs like Cursor and Claude Code. Instantly generate, search, and update code based on your API documentation — no more manual syncing or context loss.</em></blockquote><p><a href="https://roobia.medium.com/apidog-mcp-server-the-bridge-between-your-api-specifications-and-ai-coding-assistants-12d39cd44d12">Apidog MCP Server: The Bridge Between Your API Specifications and AI Coding Assistants</a></p><h3>Why Kimi K2 + Claude Code Is a Game Changer for AI Development</h3><p>Kimi K2, from Moonshot AI, is not just another large language model. Built on a Mixture-of-Experts (MoE) architecture, it activates only 32 billion of its 1 trillion parameters per task — delivering massive performance with impressive efficiency. Trained on 15.5 trillion tokens, Kimi K2 is engineered for deep reasoning, robust tool use, and, most importantly, agentic coding.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*jzASBaTMs0Vwfy8I.png" /></figure><p><strong>What’s agentic coding?</strong><br>It’s the ability for an AI to act like a true software engineer: understanding problems, planning solutions, using tools, and autonomously solving complex tasks from start to finish. Kimi K2’s benchmark results in this area are nothing short of remarkable.</p><h3>Kimi K2’s Agentic Coding Benchmarks: Setting a New Standard</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*SPvkvOEmKD998sx0.png" /></figure><ul><li><strong>SWE-bench Verified (Agentic Coding):</strong><br>Kimi K2 scores <strong>65.8%</strong> on real-world GitHub issue resolution (single attempt), outpacing GPT-4.1 and rivaling Claude 4 Opus. With multiple attempts, it hits <strong>71.6%</strong>.</li><li><strong>SWE-bench Multilingual:</strong><br>Kimi K2 achieves <strong>47.3%</strong>, far ahead of other open-source models, proving its language-agnostic coding prowess.</li><li><strong>TerminalBench &amp; Aider-Polyglot:</strong><br>Kimi K2 excels in terminal operations and multi-language development, demonstrating practical, hands-on coding ability.</li></ul><h3>The Open-Source Advantage</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Ipyq3UgeNjUKpc0e.png" /></figure><p>Kimi K2 is fully open source under a Modified MIT License. This means:</p><ul><li><strong>Transparency:</strong> Inspect and trust the model’s architecture and training.</li><li><strong>Customization:</strong> Fine-tune on your own data for specialized use cases.</li><li><strong>Democratization:</strong> Anyone with the resources can run state-of-the-art AI — no vendor lock-in.</li></ul><p>This open approach, combined with agentic coding power, makes Kimi K2 a foundational tool for the future of software development.</p><h3>Step-by-Step: How to Use Kimi K2 as the Engine for Claude Code</h3><p>The secret to this integration is Moonshot AI’s Anthropic-compatible API endpoint. By redirecting Claude Code’s requests to Kimi K2, you get all the benefits of Claude’s interface with the muscle of Kimi K2 under the hood.</p><h4>1. Register on the Moonshot AI Platform</h4><ul><li>Go to <a href="https://platform.moonshot.ai/">https://platform.moonshot.ai/</a></li><li>Sign up for an account (email and password required)</li><li>This is where you’ll manage your API keys and usage</li></ul><h4>2. Generate Your API Key</h4><ul><li>In your Moonshot AI dashboard, find the “API Keys” section</li><li>Create a new key (it will start with sk-)</li><li><strong>Copy and save this key securely</strong> — you won’t be able to view it again</li></ul><h4>3. Set Up Your Environment Variables</h4><p>Open your terminal and run:</p><pre>export ANTHROPIC_AUTH_TOKEN=sk-YOURKEY<br>export ANTHROPIC_BASE_URL=https://api.moonshot.ai/anthropic</pre><blockquote><em>On Windows, use </em><em>set instead of </em><em>export:</em></blockquote><pre>set ANTHROPIC_AUTH_TOKEN=sk-YOURKEY<br>set ANTHROPIC_BASE_URL=https://api.moonshot.ai/anthropic</pre><p>This tells Claude Code to use your Moonshot API key and to send requests to Kimi K2’s endpoint instead of Anthropic’s.</p><h4>4. Launch Claude Code</h4><p>Start Claude Code as usual. You won’t see any special notifications, but all your prompts and code requests will now be handled by Kimi K2 — right inside the familiar Claude interface.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*6ra8xA9c6zLlA26x.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*_z4rQBCkqp-aETbX.png" /></figure><h3>Why This Matters: The Future of AI Coding Is Open, Flexible, and Interoperable</h3><p>This setup isn’t just a clever hack — it’s a glimpse into the future of AI development. By enabling API compatibility, Moonshot AI empowers developers to choose the best models and interfaces for their needs, without being locked into a single ecosystem.</p><p>With Kimi K2’s open-source agentic intelligence and Claude Code’s developer-friendly environment, you get the best of both worlds: raw power and refined usability.</p><h3>Conclusion: Build Your Ultimate AI Coding Stack</h3><p>The combination of Kimi K2 and Claude Code is more than just a technical curiosity — it’s a paradigm shift. It proves that the future of AI development is about interoperability, openness, and empowering developers to build their own ideal workflows.</p><p>By following these steps, you can harness the full power of Kimi K2 inside Claude Code, and with tools like Apidog MCP Server, you can connect your API documentation directly to your AI assistant for even greater productivity.</p><p>Ready to experience the next evolution of AI-powered coding? Set up your environment, connect your favorite models, and start building the future — today.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=22b743b69e5a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Unlock Google Veo 3 for Free: 3 Proven Methods]]></title>
            <link>https://medium.com/@Erik_Milosevic/unlock-google-veo-3-for-free-3-proven-methods-edaeb6b5ed28?source=rss-f4e8d1747a77------2</link>
            <guid isPermaLink="false">https://medium.com/p/edaeb6b5ed28</guid>
            <category><![CDATA[text-to-video-ai]]></category>
            <category><![CDATA[google-veo-3]]></category>
            <category><![CDATA[google-veo-ai]]></category>
            <category><![CDATA[google-veo]]></category>
            <category><![CDATA[text-to-video-tool]]></category>
            <dc:creator><![CDATA[Erik Milošević]]></dc:creator>
            <pubDate>Tue, 19 Aug 2025 05:52:58 GMT</pubDate>
            <atom:updated>2025-08-19T05:52:58.544Z</atom:updated>
            <content:encoded><![CDATA[<p>Ever feel like the coolest AI tools are always just out of reach — locked behind a paywall, taunting you with their premium features? Google Veo 3 is the latest must-try for video generation, but $19.99/month isn’t everyone’s idea of fun. Good news: you don’t have to pay a cent if you know the right tricks. Here’s how to get Google Veo 3 for free — and how to test its API like a true developer.</p><h3>Why Everyone Wants Free Google Veo 3 (and How to Actually Get It)</h3><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*8rZ4SjUMYBgW5rBs.png" /></figure><p><a href="https://deepmind.google/models/veo/">Google Veo 3</a> isn’t just another AI toy — it’s a creative powerhouse for:</p><ul><li>Generating high-quality videos</li><li>Analyzing textbooks</li><li>Turning course materials into podcasts</li></ul><p>But with a price tag, it’s no wonder everyone’s searching for a free pass. Here’s the cheat sheet:</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*DxfidmzCqjtUEhEM.png" /></figure><h3>1. Student Promo: The 15-Month Free Ride</h3><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*mGvQ6cTnJz7nGa66.png" /></figure><p><strong>Delve into the student advantage:</strong> If you’re a university student (or can get a student email), you can indulge in 15 months of free Google Veo 3. Here’s the play-by-play:</p><h4>Step 1: Get a Student Email</h4><ul><li>Use your real university email, or generate a temporary one at <a href="https://tempemail.id/">tempemail.id</a> (verify you’re human, grab the code).</li></ul><h4>Step 2: Sign Up for the Student Promo</h4><ul><li>Head to the <a href="https://one.google.com/explore-plan/ai-premium-studen">Google One Student page</a> and start the verification.</li><li>Paste your student email, complete the process.</li><li>If you’re in Brazil, Indonesia, Japan, the UK, or the US, you’re golden. (VPN if needed!)</li></ul><h4>Step 3: Enjoy the Spoils</h4><ul><li>15 months of Veo 3, Gemini Pro, NotebookLM Pro, and more.</li><li><strong>Tip:</strong> Re-verify before August 2025 to keep the party going.</li></ul><p><strong>Heads up:</strong> Offer expires June 30, 2025. Payment method required.</p><h3>2. Google AI Pro Free Trial: The One-Month Power User Pass</h3><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*qDocIsrF9hhs1HmK.png" /></figure><p><strong>Indulge in a free month:</strong> Not a student? No problem. Anyone can get a month of Veo 3 for free with the Google AI Pro trial.</p><h4>Step 1: Start Your Free Trial</h4><ul><li>Go to <a href="https://one.google.com/about/google-ai-plans/">Google AI Plans</a> and sign up.</li><li>Enjoy 1 month of full access (then $19.99/mo if you stick around).</li></ul><h4>Step 2: Unlock the Goodies</h4><ul><li>Use Veo 3 for video generation (limited access)</li><li>Try Gemini Pro, Deep Research, NotebookLM</li><li>Upload and analyze big files (up to 1,500 pages)</li><li>2TB storage and premium perks</li></ul><p><strong>Pro Tip:</strong> Use this month to test everything, generate videos, and explore the Veo 3 API.</p><h3>3. Google Cloud $300 Free Credit: The Developer’s Playground</h3><p><strong>Delve into the developer’s route:</strong> Want to use the Google Veo 3 API directly? Google Cloud gives new users $300 in free credits. Here’s how to get started:</p><h4>Step 1: Set Up Google Cloud and Vertex AI</h4><ul><li>Sign up for a <a href="https://console.cloud.google.com/freetrial">Google Cloud account</a> and enable billing (no auto-charges).</li><li>Create a new project and enable the Vertex AI API.</li></ul><h4>Step 2: Authenticate and Set Up</h4><ul><li>Install the <a href="https://cloud.google.com/sdk/docs/install">Google Cloud CLI</a> and run gcloud init.</li><li>If you use an external IdP, <a href="https://cloud.google.com/iam/docs/workforce-log-in-gcloud">sign in with your federated identity</a>.</li></ul><h4>Step 3: Generate Videos from Text</h4><ul><li>Go to <a href="https://console.cloud.google.com/vertex-ai/studio/media/generate;tab=video">Vertex AI Studio &gt; Media Studio</a>.</li><li>Configure your settings (model, aspect ratio, video length, etc.).</li><li>Enter your prompt and hit <strong>Generate</strong>.</li></ul><p><strong>Pro Tip:</strong> Use your $300 credit for API calls and video generation. <a href="https://cloud.google.com/vertex-ai/generative-ai/docs/video/generate-videos#googlegenaisdk_videogen_with_txt-console">Official docs here.</a></p><h3>How to Test and Document the Google Veo 3 API Like a Pro (with Apidog)</h3><p>In the rapidly changing world of API development, testing and documenting your API is as important as using it. This is where <a href="https://bit.ly/3Teeyxv"><strong>Apidog</strong></a><strong> </strong>shines as the all-in-one API platform for modern teams.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*MwZpe9sYh9vq4duZ.png" /></figure><h3>Why Use Apidog for Vertex AI API with Free Google Veo 3?</h3><ul><li><strong>Unified platform:</strong> Design, test, document, and manage APIs in one place</li><li><strong>Fast Request:</strong> Instantly send API requests and get response results</li><li><strong>Automated validation:</strong> Ensure your API responses match the spec</li><li><strong>Mock data generation:</strong> Test your frontend before the backend is ready</li><li><strong>Collaboration:</strong> Share API docs and test cases with your team</li></ul><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*F75Jd_3C-ZGt8T91.png" /></figure><h3>Step-by-Step: Using Apidog for Vertex AI API with Free Google Veo 3</h3><p><strong>1. Sign Up for Apidog</strong></p><ul><li><a href="https://bit.ly/4e0MUfo">Create a free account</a>.</li></ul><p><strong>2. Import the Google Veo 3 API Spec</strong></p><ul><li>Import OpenAPI/Swagger spec or use the visual editor.</li></ul><p><strong>3. Send Requests Immediately</strong></p><ul><li>Select the endpoint, fill in parameters, and hit “Send.”</li></ul><p><strong>4. Validate and Document</strong></p><ul><li>Automated validation, assertions, and one-click docs.</li></ul><p><strong>Pro Tip:</strong> With <a href="https://bit.ly/41ItNSS">Apidog MCP Server</a>, connect your API specs to AI-powered IDEs like Cursor, letting your AI assistant generate code, search docs, and more — all based on your actual API.</p><h3>Conclusion: Don’t Just Use Veo 3 — Master It</h3><p>In the fast-moving world of AI, Google Veo 3 is a must-try for creators, students, and developers. Whether you’re leveraging a student promo, a free trial, or Google Cloud credits, there’s a way for everyone to experience Free Google Veo 3.</p><p>But don’t stop at just using the tool — test, document, and automate the Google Veo 3 API with Apidog, the all-in-one platform that takes your API workflow to the next level.</p><p><strong>Key Takeaways:</strong></p><ul><li>Three proven ways to get Free Google Veo 3: student promo, free trial, or Google Cloud credits</li><li>Use Apidog to test, document, and automate the Google Veo 3 API</li><li>Apidog MCP Server makes API development faster, smarter, and more collaborative</li></ul><p><em>Ready to unlock the full power of Google Veo 3 and your APIs? </em><a href="https://bit.ly/3Teeyxv"><em>Sign up for Apidog</em></a><em> today and experience the future of API development.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=edaeb6b5ed28" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[15 Free API Documentation Tools That Won’t Make You Cry]]></title>
            <link>https://medium.com/@Erik_Milosevic/15-free-api-documentation-tools-that-wont-make-you-cry-d870623c7b8a?source=rss-f4e8d1747a77------2</link>
            <guid isPermaLink="false">https://medium.com/p/d870623c7b8a</guid>
            <category><![CDATA[rest-api-design]]></category>
            <category><![CDATA[api-documentation-tool]]></category>
            <category><![CDATA[api-documentation]]></category>
            <category><![CDATA[api-design]]></category>
            <category><![CDATA[api-document]]></category>
            <dc:creator><![CDATA[Erik Milošević]]></dc:creator>
            <pubDate>Tue, 19 Aug 2025 04:30:20 GMT</pubDate>
            <atom:updated>2025-08-19T04:30:20.418Z</atom:updated>
            <content:encoded><![CDATA[<p>Let’s be real: writing API documentation is about as fun as debugging a production outage at 3 am. But it doesn’t have to be! There are more free tools than ever to help you create docs that are clear, interactive, and (dare we say) actually enjoyable to use.</p><p>Whether you’re a solo dev, a startup, or wrangling a massive enterprise API, this guide will help you find the right tool to turn your API docs from a source of pain into a source of pride.</p><h3>Why Good API Docs Matter (and Why Bad Docs Are a Nightmare)</h3><ul><li><strong>Faster Onboarding:</strong> Good docs = less time explaining, more time building.</li><li><strong>Fewer Support Tickets:</strong> Let your docs answer the easy stuff so your team can focus on the hard bugs.</li><li><strong>Better Collaboration:</strong> Docs are the single source of truth for your whole squad — devs, QA, PMs, and even that one guy who only speaks in Jira tickets.</li><li><strong>More API Adoption:</strong> If your docs are a mess, nobody wants to use your API. If they’re great, you’ll have fans for life.</li><li><strong>Happier Developers:</strong> Because nobody wants to read (or write) a 200-page PDF ever again.</li></ul><h3>What Makes a Great API Doc Tool?</h3><ul><li><strong>Easy to Use:</strong> If you need a PhD to get started, it’s not the tool for you.</li><li><strong>Automation:</strong> Auto-generate docs from OpenAPI/Swagger? Yes, please.</li><li><strong>Interactivity:</strong> Let users try endpoints right in the docs.</li><li><strong>Customization:</strong> Make it look like <em>your</em> brand, not a generic template.</li><li><strong>Collaboration:</strong> Real-time editing, comments, and version control are a must.</li><li><strong>Versioning:</strong> Support for multiple API versions (because breaking changes happen).</li><li><strong>Import/Export:</strong> Play nice with other formats and tools.</li></ul><h3>The 15 Best Free API Documentation Tools for 2025</h3><h4>1. Apidog — The All-in-One API Doc Platform</h4><p><a href="https://bit.ly/3Teeyxv">Apidog</a> isn’t just a doc tool — it’s your API’s best friend. Design, test, document, mock, and collaborate, all in one place. No more juggling Postman, Swagger, and a dozen browser tabs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*KtAGCfXo6r2rxDo7.png" /></figure><p><strong>Why Devs Love It:</strong></p><ul><li><strong>Instant Testing:</strong> Test endpoints as you write docs.</li><li><strong>One-Click Online Docs:</strong> Publish and share interactive docs instantly.</li><li><strong>Mock APIs:</strong> Auto-generate mock servers for painless frontend-backend handoffs.</li><li><strong>Code Generation:</strong> Export ready-to-use code for your stack.</li><li><strong>Real-Time Collaboration:</strong> Edit, comment, and version-control with your team.</li><li><strong>Visual Dashboard:</strong> No learning curve — just get building.</li></ul><p><a href="https://bit.ly/4e0MUfo">Sign up for Apidog for free</a> and see why it’s the API doc tool everyone’s talking about.</p><h4>2. Swagger UI: The OG of API Docs</h4><p><a href="https://swagger.io/tools/swagger-ui/">Swagger UI</a> is the classic, open-source tool for turning OpenAPI specs into beautiful, interactive docs. If you’ve ever seen a “Try it out” button in API docs, you’ve probably used Swagger UI.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*f3mVhuEtCYi0O58F.png" /></figure><p><strong>Why It Rocks:</strong></p><ul><li><strong>Visual, Interactive Docs:</strong> See every endpoint, parameter, and response in a clean UI.</li><li><strong>Test in the Browser:</strong> Try out API calls without leaving the docs.</li><li><strong>Auto-Updates:</strong> Docs stay in sync with your OpenAPI spec.</li><li><strong>Open Source:</strong> Free forever, with a huge community.</li></ul><h4>3. API Blueprint: Markdown for API Nerds</h4><p><a href="https://apiblueprint.org/">API Blueprint</a> lets you write API docs in Markdown, making it super readable and easy to maintain. It’s perfect for devs who want docs as code.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*9PhbxZhG1Bau0ylB.png" /></figure><p><strong>Why It’s Cool:</strong></p><ul><li><strong>Markdown Syntax:</strong> Write docs like you write README files.</li><li><strong>MSON for Data Structures:</strong> Describe complex payloads with ease.</li><li><strong>Great for Collaboration:</strong> Easy for teams to edit and review.</li><li><strong>Tons of Integrations:</strong> Tools like Aglio and Drafter make it even more powerful.</li></ul><h4>4. apiDoc — Inline Docs for RESTful APIs</h4><p><a href="https://apidocjs.com/">apiDoc</a> generates docs straight from your code comments. If you love “docs as code,” this is your jam.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*1meIIZLyg3wH7y1g.png" /></figure><p><strong>Why Devs Use It:</strong></p><ul><li><strong>Inline Annotations:</strong> Docs stay up-to-date with your code.</li><li><strong>Customizable Templates:</strong> Make your docs look however you want.</li><li><strong>Multiple Output Formats:</strong> HTML, Markdown, PDF — you name it.</li><li><strong>Versioning:</strong> Track changes and compare API versions.</li></ul><h4>5. Apiary — Interactive Docs for Teams</h4><p><a href="https://apiary.io/">Apiary</a> is all about making API docs interactive, collaborative, and easy to use. It’s built for teams who want to design, mock, and document APIs together.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*25m5HPJmqEuCU0vS.png" /></figure><p><strong>Why Teams Love It:</strong></p><ul><li><strong>Interactive Docs:</strong> Explore endpoints and try requests in the browser.</li><li><strong>Mock Servers:</strong> Test your API before you write a single line of backend code.</li><li><strong>Rapid Prototyping:</strong> Validate designs early and avoid integration nightmares.</li><li><strong>Code Generation:</strong> Get client code in your favorite language.</li></ul><h4>6. DapperDox — Docs That Don’t Suck</h4><p><a href="http://dapperdox.io/">DapperDox</a> turns your OpenAPI specs and Markdown guides into beautiful, browsable reference sites. It’s open source and super customizable.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*_CPjUcaOfvBpz-QR.png" /></figure><p><strong>Why It’s Different:</strong></p><ul><li><strong>Markdown + OpenAPI:</strong> Mix rich guides with your API reference.</li><li><strong>Multiple Specs:</strong> Document big, complex APIs with ease.</li><li><strong>API Explorer:</strong> Test endpoints right from the docs.</li><li><strong>Custom Themes:</strong> Make it look like your brand.</li></ul><h4>7. DocFX — .NET Devs, Rejoice!</h4><p><a href="https://dotnet.github.io/docfx/">DocFX</a> is the go-to for .NET projects, but it works for any language. Generate static sites from code comments and Markdown.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/912/0*PdzcnphVEspN07wW.png" /></figure><p><strong>Why It’s Handy:</strong></p><ul><li><strong>API Reference from Code:</strong> Auto-generate docs from XML comments.</li><li><strong>Static Site Generator:</strong> Host anywhere, no server needed.</li><li><strong>Cross-Platform:</strong> Works on Windows, Mac, and Linux.</li><li><strong>Customizable:</strong> Templates and plugins galore.</li></ul><h4>8. Document360 — Docs for Grown-Ups</h4><p><a href="https://document360.com/">Document360</a> is a flexible platform for building, managing, and publishing API docs. It’s great for teams who want a polished, interactive experience.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*tHMgV8jBYouRajWh.png" /></figure><p><strong>Why It’s Great:</strong></p><ul><li><strong>Auto-Gen from OpenAPI:</strong> Save time and avoid manual errors.</li><li><strong>Interactive “Try It” Feature:</strong> Let users test endpoints live.</li><li><strong>Customizable &amp; Versioned:</strong> Make it yours and manage multiple API versions.</li><li><strong>Comprehensive Coverage:</strong> Every endpoint, parameter, and model covered.</li></ul><h4>9. Doxygen — The Old-School Powerhouse</h4><p><a href="https://www.doxygen.nl/">Doxygen</a> is the granddaddy of code documentation tools. It’s not just for APIs — use it for any codebase, in almost any language.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*D0CGZxlTKXJeVzYG.png" /></figure><p><strong>Why It’s Still Around:</strong></p><ul><li><strong>Multi-Language Support:</strong> C++, Python, Java, PHP, and more.</li><li><strong>Cross-References &amp; Diagrams:</strong> Navigate huge codebases with ease.</li><li><strong>Multiple Output Formats:</strong> HTML, PDF, Word, XML — you name it.</li><li><strong>Automates the Boring Stuff:</strong> Docs from code comments, no sweat.</li></ul><h4>10. GitBook — Docs That Look Like a Million Bucks</h4><p><a href="https://www.gitbook.com/">GitBook</a> is perfect for teams who want beautiful, collaborative docs with real-time editing and Git integration.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*nxG74B8Tw8v1PgQl.png" /></figure><p><strong>Why It’s a Hit:</strong></p><ul><li><strong>User-Friendly Editor:</strong> Write, edit, and organize docs with zero friction.</li><li><strong>Real-Time Collaboration:</strong> Multiple people can work together, Google Docs-style.</li><li><strong>Git Integration:</strong> Track changes and roll back when needed.</li><li><strong>Markdown Support:</strong> Keep it simple and consistent.</li></ul><h4>11. OpenAPI Generator — Code &amp; Docs, All in One</h4><p><a href="https://openapi-generator.tech/">OpenAPI Generator</a> isn’t just for docs — it can generate client libraries, server stubs, and interactive docs from your OpenAPI spec.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Z6jGthqBVGWDZJP1.png" /></figure><p><strong>Why It’s a Game-Changer:</strong></p><ul><li><strong>Code Generation:</strong> Get client/server code for dozens of languages.</li><li><strong>Interactive Docs:</strong> Explore and test APIs right from the docs.</li><li><strong>API-First:</strong> Define your spec, generate everything else.</li></ul><h4>12. Postman — Not Just for Testing</h4><p><a href="https://www.postman.com/">Postman</a> is famous for API testing, but its documentation features are top-notch too.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*U4rRhlPG6ioAg-tA.png" /></figure><p><strong>Why It’s Useful:</strong></p><ul><li><strong>Auto-Gen Docs:</strong> Collections turn into docs automatically.</li><li><strong>Interactive:</strong> Test endpoints from the docs.</li><li><strong>Collaboration:</strong> Share, comment, and update with your team.</li><li><strong>Customizable:</strong> Add descriptions, examples, and more.</li></ul><h4>13. RAML — Docs That Stay in Sync</h4><p><a href="https://raml.org/developers/document-your-api">RAML</a> is all about keeping your docs and your API in perfect harmony. Define your API once, and generate docs, consoles, and more.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*5FDPmb4Ourd36zo-.png" /></figure><p><strong>Why It’s Smart:</strong></p><ul><li><strong>API Console:</strong> Live, interactive docs for your users.</li><li><strong>RAML to HTML:</strong> Generate beautiful docs in seconds.</li><li><strong>Always in Sync:</strong> No more drift between code and docs.</li></ul><h4>14. ReadMe — Docs That Don’t Suck</h4><p><a href="https://readme.com/">ReadMe</a> is a robust platform for creating, managing, and distributing API docs. It’s all about making docs that are actually helpful.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*zPm8xVRiFh05z7Zn.png" /></figure><p><strong>Why It’s Popular:</strong></p><ul><li><strong>Project Overview:</strong> Give users the big picture.</li><li><strong>Easy Setup:</strong> Installation, usage, and contribution guides built in.</li><li><strong>Collaboration:</strong> Get your whole team involved.</li><li><strong>Clarity:</strong> No more confusing, outdated docs.</li></ul><h4>15. Redoc — Open Source, OpenAPI Magic</h4><p><a href="https://github.com/Redocly/redoc">Redoc</a> is an open-source tool that turns your OpenAPI specs into gorgeous, interactive docs with a modern three-panel layout.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*e-UxsPGV6TibzAUs.png" /></figure><p><strong>Why It’s a Favorite:</strong></p><ul><li><strong>Automatic Generation:</strong> Docs update with your OpenAPI spec.</li><li><strong>Three-Panel Layout:</strong> Navigation, details, and examples all in one view.</li><li><strong>Customizable:</strong> Make it match your brand.</li><li><strong>Open Source:</strong> Free and community-driven.</li><li><strong>Interactive:</strong> Explore endpoints and responses with ease.</li></ul><h3>Conclusion: Make Your API Docs a Superpower, Not a Slog</h3><p>Great API docs are the secret weapon of every successful dev team. With these 15 free tools, you can finally ditch the clunky PDFs and confusing wikis for docs that are interactive, up-to-date, and actually fun to use.</p><p>If you want to go from “meh” to “amazing” in your API workflow, give Apidog a try — it’s the all-in-one platform that makes API docs (and everything else) a breeze.</p><p>Explore these tools, experiment, and find the one that fits your team’s vibe. Your future self (and your users) will thank you!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d870623c7b8a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[10 Must-Try Small Local LLMs That Run on Less Than 8GB RAM/VRAM]]></title>
            <link>https://medium.com/@Erik_Milosevic/10-must-try-small-local-llms-that-run-on-less-than-8gb-ram-vram-fdc514c51140?source=rss-f4e8d1747a77------2</link>
            <guid isPermaLink="false">https://medium.com/p/fdc514c51140</guid>
            <category><![CDATA[local-llm]]></category>
            <category><![CDATA[llm]]></category>
            <category><![CDATA[llm-agent]]></category>
            <category><![CDATA[llm-applications]]></category>
            <dc:creator><![CDATA[Erik Milošević]]></dc:creator>
            <pubDate>Tue, 19 Aug 2025 03:27:19 GMT</pubDate>
            <atom:updated>2025-08-19T03:27:19.159Z</atom:updated>
            <content:encoded><![CDATA[<p>When most people think of Large Language Models (LLMs), they picture massive cloud servers and hefty subscription fees. But the AI revolution is now at your fingertips — literally. Thanks to cutting-edge quantization and model optimization, you can run powerful LLMs right on your laptop or desktop, even if you have less than 8GB of RAM or VRAM. Let’s explore how you can bring advanced AI to your local machine, and which models are leading the charge.</p><blockquote><strong><em>Pro Tip:</em></strong><em> </em><strong><em>Did you know? </em></strong><a href="https://bit.ly/3Teeyxv"><strong><em>Apidog</em></strong></a><strong><em> is the all-in-one API development platform that lets you design, test, and document APIs with ease — no matter if you’re working with cloud or local AI models. </em></strong><a href="https://bit.ly/4e0MUfo"><strong><em>Try Apidog for free</em></strong></a><strong><em> and experience seamless integration with your favorite tools!</em></strong></blockquote><h3>Demystifying Quantization: How Small LLMs Fit on Modest Hardware</h3><p>Before diving into the best models, let’s break down the tech that makes local LLMs possible. The secret sauce? Quantization — a process that shrinks model weights from 16- or 32-bit floats to 4- or 8-bit integers, slashing memory requirements without a major hit to quality. For example, a 7B parameter model that would normally need 14GB in FP16 can run in just 4–5GB with 4-bit quantization.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*z4n_AvUFQGa9Xvti.png" /></figure><p><strong>Key concepts:</strong></p><ul><li><strong>VRAM vs. RAM:</strong> VRAM (on your GPU) is fast and ideal for LLM inference; RAM (system memory) is slower but more plentiful. For best results, keep the model in VRAM.</li><li><strong>GGUF Format:</strong> The go-to format for quantized models, compatible with most local inference engines.</li><li><strong>Quantization Types:</strong> Q4_K_M is a sweet spot for quality and efficiency; Q2_K or IQ3_XS save more space but may reduce output quality.</li><li><strong>Memory Overhead:</strong> Always budget about 1.2x the model file size to account for activations and prompt context.</li></ul><h3>Getting Started: Tools for Running Local LLMs</h3><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*CYCTE8xabLFIF2UA.png" /></figure><ul><li><strong>Ollama:</strong> A developer-friendly CLI tool for running LLMs locally. It’s fast, scriptable, and supports custom model packaging via Modelfile. Perfect for coders and automation pros.</li><li><strong>LM Studio:</strong> Prefer a GUI? LM Studio offers a slick desktop app with built-in chat, easy model downloads from Hugging Face, and simple parameter tweaking. Great for beginners and non-coders.</li><li><strong>Llama.cpp:</strong> The C++ engine powering many local LLM tools, optimized for GGUF models and both CPU/GPU acceleration.</li></ul><h3>The Top 10 Small Local LLMs (All Under 8GB!)</h3><h4>1. Llama 3.1 8B (Quantized)</h4><pre>ollama run llama3.1:8b</pre><p>Meta’s Llama 3.1 8B is a standout for general-purpose AI, boasting a huge training set and smart optimizations. Quantized versions like Q2_K (3.18GB file, ~7.2GB memory) and Q3_K_M (4.02GB file, ~7.98GB memory) make it accessible for most laptops. It shines in chat, code, summarization, and RAG tasks, and is a favorite for batch processing and agentic workflows.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*OjtCPzejr0r4vMpH.png" /></figure><h4>2. Mistral 7B (Quantized)</h4><pre>ollama run mistral:7b</pre><p>Mistral 7B is engineered for speed and efficiency, using GQA and SWA for top-tier performance. Q4_K_M (4.37GB file, 6.87GB memory) and Q5_K_M (5.13GB file, 7.63GB memory) quantizations are perfect for 8GB setups. It’s ideal for real-time chatbots, edge devices, and commercial use (Apache 2.0 license).</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*STiBI06RNQjE1aP4.png" /></figure><h4>3. Gemma 3:4B (Quantized)</h4><pre>ollama run gemma3:4b</pre><p>Google DeepMind’s Gemma 3:4B is tiny but mighty. Q4_K_M (1.71GB file) runs on just 4GB VRAM, making it perfect for mobile and low-end PCs. Great for text generation, Q&amp;A, and OCR tasks.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*jXNf8S-yt6FxnqF2.png" /></figure><h4>4. Gemma 7B (Quantized)</h4><pre>ollama run gemma:7b</pre><p>The bigger Gemma 7B brings more muscle for code, math, and reasoning, but still fits in 8GB VRAM (Q5_K_M: 6.14GB, Q6_K: 7.01GB). It’s versatile for content creation, chat, and knowledge work.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*QFGK3ELMVnmB4H4P.png" /></figure><h4>5. Phi-3 Mini (3.8B, Quantized)</h4><pre>ollama run phi3</pre><p>Microsoft’s Phi-3 Mini is a compact powerhouse for logic, coding, and math. Q8_0 (4.06GB file, 7.48GB memory) is well within the 8GB limit. It’s great for chat, mobile, and latency-sensitive tasks.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*N22qH7zMg0iRCNWE.png" /></figure><h4>6. DeepSeek R1 7B/8B (Quantized)</h4><pre>ollama run deepseek-r1:7b</pre><p>DeepSeek’s 7B and 8B models are renowned for reasoning and code. The R1 7B Q4_K_M (4.22GB file, 6.72GB memory) and R1 8B (4.9GB file, 6GB VRAM) are both 8GB-friendly. They’re ideal for SMBs, customer support, and advanced data analysis.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*SrDa2b9o7pcpxTMB.png" /></figure><h4>7. Qwen 1.5/2.5 7B (Quantized)</h4><pre>ollama run qwen:7b</pre><p>Alibaba’s Qwen 7B models are multilingual and context-rich (32K tokens). Qwen 1.5 7B Q5_K_M (5.53GB) and Qwen2.5 7B (4.7GB, 6GB VRAM) are perfect for chatbots, translation, and programming help.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*pkwWD7GUPVzKKp5a.png" /></figure><h4>8. Deepseek-coder-v2 6.7B (Quantized)</h4><pre>ollama run deepseek-coder-v2:6.7b</pre><p>Deepseek-coder-v2 6.7B is a coder’s dream — fine-tuned for code generation and understanding. At 3.8GB (6GB VRAM), it’s a top pick for local code completion and developer tools.</p><h4>9. BitNet b1.58 2B4T</h4><pre>ollama run hf.co/microsoft/bitnet-b1.58-2B-4T-gguf</pre><p>BitNet b1.58 2B4T from Microsoft is a marvel of efficiency, using 1.58-bit weights to run in just 0.4GB of memory. It’s perfect for edge devices, IoT, and CPU-only inference — think on-device translation and mobile assistants.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Gri6vGUXABcH5fMz.png" /></figure><h4>10. Orca-Mini 7B (Quantized)</h4><pre>ollama run orca-mini:7b</pre><p>Orca-Mini 7B, built on Llama and Llama 2, is a flexible model for chat, Q&amp;A, and instruction following. Q4_K_M (4.08GB file, 6.58GB memory) and Q5_K_M (4.78GB file, 7.28GB memory) are both 8GB-friendly. It’s great for building AI agents and conversational tools.</p><h3>Final Thoughts: The Future of Local LLMs Is Here</h3><p>The models above — Llama 3.1 8B, Mistral 7B, Gemma 3:4B and 7B, Phi-3 Mini, DeepSeek R1, Qwen 7B, Deepseek-coder-v2, BitNet b1.58, and Orca-Mini — prove that you don’t need a supercomputer to harness AI. Thanks to quantization and open-source innovation, you can run advanced language models on everyday hardware.</p><p><strong>Why does this matter?</strong></p><ul><li><strong>Privacy:</strong> Keep your data local — no cloud required.</li><li><strong>Cost:</strong> No subscriptions or cloud fees.</li><li><strong>Speed:</strong> Instant responses, even offline.</li><li><strong>Flexibility:</strong> Experiment, customize, and deploy anywhere.</li></ul><p>As quantization and edge AI keep advancing, expect even more powerful models to run on smaller devices. Dive in, experiment, and find the LLM that fits your workflow. And if you’re building APIs or integrating AI into your stack, don’t forget to <a href="https://bit.ly/3Teeyxv"><em>check out Apidog</em></a> for a seamless, all-in-one development experience!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fdc514c51140" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Unlocking Augment Code: Free Strategies]]></title>
            <link>https://medium.com/@Erik_Milosevic/unlocking-augment-code-free-strategies-d5046a5082cf?source=rss-f4e8d1747a77------2</link>
            <guid isPermaLink="false">https://medium.com/p/d5046a5082cf</guid>
            <category><![CDATA[ai-coding]]></category>
            <category><![CDATA[ideais]]></category>
            <category><![CDATA[augment-code]]></category>
            <category><![CDATA[free-ai-tools]]></category>
            <dc:creator><![CDATA[Erik Milošević]]></dc:creator>
            <pubDate>Wed, 04 Jun 2025 09:52:13 GMT</pubDate>
            <atom:updated>2025-06-04T09:52:13.890Z</atom:updated>
            <content:encoded><![CDATA[<p>AI coding tools like <a href="https://www.augmentcode.com/">Augment Code</a> are rapidly changing how developers work. While it offers a free version and a 14-day trial, many wonder how to continue using its powerful features without a subscription. This guide reveals five clever ways to keep Augment Code at your fingertips for free, along with how <a href="https://bit.ly/3Teeyxv">Apidog</a> and its <a href="https://bit.ly/41ItNSS">Apidog MCP Server</a> can amplify its capabilities, especially for API-centric projects.</p><h3>1. Embrace Augment Code’s Complimentary Tier</h3><p>For those new to AI-assisted coding, <a href="https://www.augmentcode.com/pricing">Augment Code’s free version</a> is an excellent starting point. It provides essential features and a generous allowance of interactions (e.g., 50 messages), making it perfect for small-scale projects or educational purposes. Simply sign up, integrate it into your preferred IDE, and begin experiencing AI assistance for tasks like code generation and debugging. It’s a zero-cost entry into the world of AI-powered development.</p><h3>2. Leverage the Developer Plan’s Free Trial</h3><p>Ready to explore more advanced functionalities? The Augment Code Developer plan offers a 14-day free trial that unlocks its full suite of professional features. This is your prime opportunity to temporarily access and utilize Augment Code’s premium offerings without commitment. Install the IDE extension and rigorously test advanced chat capabilities, ‘Next Edit’ suggestions, and powerful code completions on your complex projects. This trial period is invaluable for assessing its comprehensive potential before considering a paid subscription.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*fBIPmj0X-9aq-cd0.png" /></figure><h3>3. The Multi-Account Approach (Use with Discretion)</h3><p>Some users find that creating new accounts with different email addresses (e.g., Google, Microsoft, GitHub) can extend their free access to Augment Code. While this method can provide temporary continued usage, it’s crucial to be aware of Augment Code’s terms of service. Circumventing usage limits might lead to account restrictions. Consider this a short-term workaround rather than a sustainable long-term strategy for free access.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*AQloysofxN9gktSz.png" /></figure><h3>4. Engage with the Augment Code Open-Source Initiative</h3><p>Are you an open-source contributor? <a href="https://www.augmentcode.com/opensource">Augment Code’s Open-Source Program</a> grants free access to project maintainers and significant contributors. This is a mutually beneficial way to use Augment Code for free while actively supporting the open-source community. <a href="https://auth.augmentcode.com/signup/login?us=true">Apply</a> by demonstrating your involvement in open-source projects. Be mindful that data generated through this program might be used for AI model training. This initiative helps accelerate project onboarding and assists maintainers with code reviews.</p><h3>5. Explore Community-Driven GitHub Solutions</h3><p>Developer communities frequently develop unofficial tools, such as “<a href="https://github.com/vber/free-augmentcode/blob/main/README.md">Free AugmentCode</a>” found on GitHub. These tools typically aim to reset local Augment Code data (like telemetry IDs or database entries) to facilitate fresh logins with different accounts, effectively extending free usage. Exercise caution when using such unofficial tools: always verify their source for trustworthiness and understand the inherent risks, as they may become obsolete or cause software conflicts.</p><h3>Supercharge Augment Code with Apidog and Apidog MCP Server</h3><p>Beyond simply gaining free access, the true power of AI coding tools emerges through intelligent integration. Apidog, a comprehensive API platform, and its innovative Apidog MCP Server can significantly elevate your Augment Code experience, particularly for API development workflows. Apidog streamlines the entire API lifecycle, from design to testing. The Apidog MCP Server enables AI-powered IDEs like Cursor or VS Code to directly consume your API specifications.</p><p>This seamless integration means Augment Code can fetch precise API details (endpoints, schemas, parameters) directly from your up-to-date API documentation. This drastically reduces errors, saves valuable development time, and ensures your generated code remains perfectly aligned with the latest API definitions.</p><h3>Integrating Apidog MCP Server with Your IDE</h3><p>Connecting the Apidog MCP Server to your IDE allows your AI assistant to directly access and utilize your API specifications. Here’s a simplified setup guide:</p><p><strong>Prerequisites:</strong></p><p>Before you begin, ensure the following are in place:</p><p>✅ Node.js is installed (version <strong>18+</strong>; latest LTS is recommended)</p><p>✅ You are using an IDE that supports MCP, such as: <strong>Cursor</strong></p><p><strong>Step 1: Prepare Your OpenAPI Definition</strong></p><p>You will need access to your API definition file:</p><ul><li>A <strong>URL</strong> (e.g., <a href="https://petstore.swagger.io/v2/swagger.json)">https://petstore.swagger.io/v2/swagger.json)</a></li><li>Or a <strong>local file path</strong> (e.g., ~/projects/api-docs/openapi.yaml)</li><li><strong>Supported formats</strong>: .json or .yaml (OpenAPI 3.x is highly recommended)</li></ul><p><strong>Step 2: Configure MCP in Your IDE</strong></p><p>Next, you’ll add the necessary configuration to your IDE’s mcp.json file.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*DFrfDuL-mLIYWO-G.png" /></figure><p>Remember to <strong>Replace</strong> &lt;oas-url-or-path&gt; with the actual URL or local path to your OpenAPI file.</p><ul><li>For MacOS/Linux:</li></ul><pre>{<br>  &quot;mcpServers&quot;: {<br>    &quot;API specification&quot;: {<br>      &quot;command&quot;: &quot;npx&quot;,<br>      &quot;args&quot;: [<br>        &quot;-y&quot;,<br>        &quot;apidog-mcp-server@latest&quot;,<br>        &quot;--oas=https://petstore.swagger.io/v2/swagger.json&quot;<br>      ]<br>    }<br>  }<br>}</pre><p>For Windows:</p><pre>{<br>  &quot;mcpServers&quot;: {<br>    &quot;API specification&quot;: {<br>      &quot;command&quot;: &quot;cmd&quot;,<br>      &quot;args&quot;: [<br>        &quot;/c&quot;,<br>        &quot;npx&quot;,<br>        &quot;-y&quot;,<br>        &quot;apidog-mcp-server@latest&quot;,<br>        &quot;--oas=https://petstore.swagger.io/v2/swagger.json&quot;<br>      ]<br>    }<br>  }<br>}</pre><p><strong>Step 3: Validate the Connection</strong></p><p>After saving your configuration, test the connection within your IDE by typing the following command in Agent mode:</p><pre>Please fetch API documentation via MCP and tell me how many endpoints exist in the project.</pre><p>If the setup is successful, you will receive a structured response detailing the endpoints. If not, double-check the path to your OpenAPI file and ensure Node.js is correctly installed.</p><h3>Conclusion: Smart, Accessible Coding with Augment Code and Apidog</h3><p>Numerous avenues exist to access Augment Code’s powerful features for free, ranging from its complimentary tier and trial periods to engaging with open-source programs and community-driven tools. These options collectively make AI-assisted coding more accessible to a wider audience.</p><p>However, achieving peak productivity stems from intelligent tool integration. Connecting Augment Code to your API specifications through the Apidog MCP Server represents a significant leap forward. Apidog ensures that Augment Code operates with accurate, real-time API information, thereby streamlining development workflows and drastically minimizing errors. Explore these free Augment Code methods and unlock the profound efficiency gains offered by Apidog integration. Sign up for Apidog today to transform your API-first development process.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d5046a5082cf" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Unlocking Free AI Video Creation: Your Guide to Microsoft Bing Video Creator and OpenAI Sora]]></title>
            <link>https://medium.com/@Erik_Milosevic/unlocking-free-ai-video-creation-your-guide-to-microsoft-bing-video-creator-and-openai-sora-4ad60175796c?source=rss-f4e8d1747a77------2</link>
            <guid isPermaLink="false">https://medium.com/p/4ad60175796c</guid>
            <category><![CDATA[ai-video-creator]]></category>
            <category><![CDATA[ai-video-generator]]></category>
            <category><![CDATA[openai-sora]]></category>
            <category><![CDATA[ai-video]]></category>
            <category><![CDATA[bing-video-creator]]></category>
            <dc:creator><![CDATA[Erik Milošević]]></dc:creator>
            <pubDate>Wed, 04 Jun 2025 09:43:08 GMT</pubDate>
            <atom:updated>2025-06-04T09:43:08.387Z</atom:updated>
            <content:encoded><![CDATA[<p>OpenAI’s Sora text-to-video model has revolutionized AI-generated content, producing astonishingly realistic videos from simple text prompts. While its previous subscription costs ($20–200/month) limited access for many, Microsoft’s integration of Sora into Bing has democratized this advanced technology, making it freely available. This guide will walk you through accessing and utilizing OpenAI’s Sora for free via Microsoft Bing Video Creator, comparing it with the paid version, and offering practical advice to maximize your AI video output.</p><blockquote><strong>Pro Tip:</strong> As you delve into AI-powered video creation, remember that robust API management is key for integrating these powerful tools into your workflows. <a href="https://bit.ly/3Teeyxv">Apidog</a> is your comprehensive solution for <a href="https://bit.ly/4hKOxQZ">designing</a>, <a href="https://bit.ly/3Cs56kA">documenting</a>, <a href="https://bit.ly/3CBrRlY">debugging</a>, and <a href="https://bit.ly/4fMLpCJ">testing</a> APIs, ensuring seamless interaction with services like OpenAI Sora. Streamline your development and save valuable time with Apidog’s all-in-one platform.</blockquote><h3>Elevating AI Video Development with API Integration</h3><p>While free access to Sora through Bing Video Creator is a game-changer for content creators, developers building applications around AI-generated video require dependable API management. Effective API handling is crucial when embedding video generation capabilities into custom applications.</p><p>Apidog provides developers with an efficient way to interact with AI video generation APIs. Its extensive platform enables you to test video generation endpoints with various parameters, document your implementation for team collaboration, and create mock responses during development — saving on costly API calls. Apidog’s intuitive visual interface facilitates quick troubleshooting of integration issues and monitoring API performance in production.</p><h3>The Significance of Free Sora Access</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*rWdVSB7pdK6M9Ch2.png" /></figure><p>OpenAI’s Sora generates realistic videos from text prompts using sophisticated AI that comprehends spatial relationships and motion. Previously, this powerful tool was exclusively available through paid subscriptions:</p><ul><li>ChatGPT Plus ($20/month)</li><li>ChatGPT Pro ($200/month)</li></ul><p>This pricing model restricted access to businesses and professionals, excluding:</p><ul><li>Independent creators with limited budgets</li><li>Students and educators</li><li>Small businesses</li><li>Enthusiasts eager to experiment</li></ul><p>In June 2025, Microsoft transformed the landscape by integrating Sora into Bing Video Creator, offering the technology completely free to anyone with a Microsoft account. Building on their success with DALL-E for image generation, this move democratizes professional video creation, enabling users without technical expertise or significant financial resources to produce AI-generated videos for the first time.</p><h3>Your Step-by-Step Guide to Bing Video Creator with Sora</h3><p>Accessing free Sora-powered video creation is straightforward. Here’s a comprehensive guide to getting started with Bing Video Creator:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ohA1J4DxCCgPU-e3.png" /></figure><h4>Step 1: Download the Bing Mobile App</h4><p>Currently, Bing Video Creator is exclusively available via the mobile application:</p><ol><li>Open your device’s app store (App Store for iOS or Google Play Store for Android).</li><li>Search for “Microsoft Bing.”</li><li>Download and install the official Bing app.</li><li>Launch the app after installation.</li></ol><p><strong>Note:</strong> Microsoft has announced plans to extend access to desktop users through Bing.com/create and Copilot Search in the future.</p><h4>Step 2: Sign In with a Microsoft Account</h4><p>A Microsoft account is required to use Bing Video Creator:</p><ol><li>Open the Bing app.</li><li>Tap the profile icon in the top-right corner.</li><li>Select “Sign in” and enter your Microsoft account credentials.</li><li>If you don’t have an account, select “Create one” and follow the registration process.</li></ol><p>Creating a Microsoft account is free and grants access to various Microsoft services beyond Bing Video Creator.</p><h4>Step 3: Access the Video Creator Tool</h4><p>There are two primary ways to access the video generation feature:</p><p><strong>Method 1: Via the Menu</strong></p><ol><li>Open the Bing mobile app.</li><li>Tap the menu icon in the bottom-right corner.</li><li>Select “Video Creator” from the options.</li></ol><p><strong>Method 2: Direct Search</strong></p><ol><li>In the Bing search bar, type “Create a video of…” followed by your description.</li><li>The app will automatically direct you to the Video Creator interface.</li></ol><h4>Step 4: Generate Your First AI Video</h4><p>Once you’ve accessed the Video Creator tool:</p><ol><li>Enter your text prompt in the designated field.</li><li>Review your prompt for clarity and specificity.</li><li>Tap “Generate” to initiate the creation process.</li><li>Wait for a notification that your video is ready (processing time varies).</li><li>View, download, or share your newly created video.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*CJrp9RN72dxMjbys.png" /></figure><h3>Mastering Prompts for Bing Video Creator</h3><p>The quality of AI-generated videos largely depends on the clarity and detail of your prompts. Here are strategies for crafting effective prompts that yield superior results:</p><h4>Essential Elements of an Effective Prompt</h4><p>For optimal outcomes, include these elements in your video prompts:</p><p><strong>1. Subject and Setting</strong></p><pre>A young woman in a red coat walking through a snowy forest</pre><p>(Preferable to a generic “a person walking”)</p><p><strong>2. Lighting and Atmosphere</strong></p><pre>&quot;...at sunrise with golden light filtering through the trees&quot;</pre><p><strong>3. Camera Perspective</strong></p><pre>&quot;...filmed from a low angle, tracking shot that follows her movement&quot;</pre><p><strong>4. Motion and Action</strong></p><pre>&quot;...as she carefully steps over fallen logs, her breath visible in the cold air&quot;</pre><p><strong>5. Emotional Tone</strong></p><pre>&quot;...with a contemplative expression, suggesting a personal journey&quot;</pre><p><strong>Complete Example Prompt:</strong></p><pre>A young woman in a red coat walking through a snowy forest at sunrise with golden light filtering through the trees. Filmed from a low angle, tracking shot that follows her movement as she carefully steps over fallen logs, her breath visible in the cold air. She has a contemplative expression, suggesting a personal journey.</pre><h4>Sample Prompt Templates</h4><p><strong>Nature Scene:</strong></p><pre>A [weather condition] day in a [landscape type], featuring [specific elements]. The camera [camera movement] to reveal [subject] [performing action] with [descriptive details].</pre><p><strong>Character-Based:</strong></p><pre>A [adjective describing person] wearing [clothing description] is [action verb] in a [location]. The lighting is [lighting description] creating a [mood] atmosphere.</pre><p><strong>Product Showcase:</strong></p><pre>A [product] displayed on a [surface type] with [background elements]. The camera [movement type] to highlight the [product feature] as [environmental effect] occurs.</pre><h3>Tips for Enhanced Results</h3><ul><li><strong>Be Specific</strong>: Include precise details like colors, materials, and environmental elements.</li><li><strong>Avoid Ambiguity</strong>: Clear descriptions prevent misinterpretations by the AI.</li><li><strong>Limit Complexity</strong>: While Sora is advanced, simpler scenes often render more reliably.</li><li><strong>Consider Limitations</strong>: Keep the 5-second duration in mind when planning your scene.</li><li><strong>Iterate and Refine</strong>: Use initial results to guide improvements in subsequent prompts.</li></ul><h3>Understanding Bing Video Creator: Features and Limitations</h3><p>While Bing Video Creator offers impressive free capabilities, it’s important to understand its current specifications and limitations:</p><h4>Video Specifications</h4><ul><li><strong>Duration</strong>: 5 seconds per video</li><li><strong>Aspect Ratio</strong>: 9:16 vertical format (optimized for social media)</li><li><strong>Resolution</strong>: Mobile-optimized quality</li><li><strong>Format</strong>: MP4 video files</li><li><strong>Storage</strong>: Videos are stored on Microsoft servers for 90 days</li></ul><h4>Usage Limits and Processing Time</h4><p>Bing Video Creator employs a tiered approach to video generation:</p><p><strong>Fast Generation:</strong></p><ul><li>10 free “Fast” generations are provided initially.</li><li>Additional Fast generations cost 100 Microsoft Rewards points each.</li><li>Processing typically completes within a few minutes.</li></ul><p><strong>Standard Generation:</strong></p><ul><li>Unlimited free Standard generations.</li><li>Longer processing times (ranging from 30 minutes to several hours).</li><li>Processing time depends on prompt complexity and system load.</li></ul><h4>Microsoft Rewards Integration</h4><p>To encourage platform engagement, Microsoft has integrated Video Creator with its Rewards program:</p><ul><li>Earn 5 points per PC search using Bing (daily cap of 150 points).</li><li>Each Fast video generation costs 100 Rewards points.</li><li>Points can also be earned through Microsoft Store purchases and other Microsoft services.</li><li>There is no direct monetary payment option for additional Fast generations.</li></ul><h3>Free vs. Paid Sora Access: A Comparative Analysis</h3><p>Both the free Bing implementation and the paid OpenAI subscription utilize the same underlying Sora model, but they differ significantly in features and capabilities:</p><h4>Feature Comparison Table</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3a6WBB85-b2RA0omGJggvg.png" /></figure><h3>Who Benefits from Each Version?</h3><p><strong>Bing Video Creator (Free) is ideal for:</strong></p><ul><li>Social media content creators focused on short-form content.</li><li>Students and educators exploring AI technology.</li><li>Casual users experimenting with AI video generation.</li><li>Small businesses with limited content creation budgets.</li></ul><p><strong>OpenAI Sora (Paid) is better suited for:</strong></p><ul><li>Professional content creators and marketers.</li><li>Users requiring longer videos with higher resolution.</li><li>Projects demanding advanced editing capabilities.</li><li>Teams needing consistent, faster processing times.</li><li>Users who require expanded creative control options.</li></ul><h3>Practical Applications of Free AI Video Creation</h3><p>The availability of free AI video generation opens up numerous possibilities across various use cases:</p><h3>Social Media Content</h3><p><strong>TikTok and Instagram Reels:</strong></p><ul><li>Create captivating introductions for longer content.</li><li>Generate visual backgrounds for quotes or statistics.</li><li>Develop creative transitions between clips.</li><li>Visualize concepts difficult to film traditionally.</li></ul><p><strong>Example Prompt:</strong></p><pre>A smartphone floating in space showing social media notifications popping up in vibrant colors, with digital particles swirling around it, viewed from a rotating camera angle</pre><h3>Education and Explanation</h3><p><strong>Visual Learning Aids:</strong></p><ul><li>Illustrate scientific concepts with dynamic visualizations.</li><li>Create historical scene recreations.</li><li>Demonstrate physical processes.</li><li>Visualize abstract mathematical concepts.</li></ul><p><strong>Example Prompt:</strong></p><pre>A cross-section view of a plant cell with clearly visible organelles highlighted in different colors, zooming slowly from whole cell to detailed view of the mitochondria</pre><h3>Marketing and Promotion</h3><p><strong>Product Showcases:</strong></p><ul><li>Generate product demonstrations in diverse environments.</li><li>Create seasonal promotional content.</li><li>Visualize before-and-after scenarios.</li><li>Develop eye-catching ad concepts.</li></ul><p><strong>Example Prompt:</strong></p><pre>A sleek black coffee maker on a modern kitchen counter with morning sunlight streaming in. Steam rises from a freshly brewed cup as a person&#39;s hand reaches into frame.</pre><h3>Creative Storytelling</h3><p><strong>Concept Development:</strong></p><ul><li>Visualize character designs and settings.</li><li>Create mood boards for larger projects.</li><li>Test visual storytelling approaches.</li><li>Develop creative writing prompts.</li></ul><p><strong>Example Prompt:</strong></p><pre>A mysterious figure in a long coat standing at the edge of a misty forest at dusk, their face partially illuminated by the warm glow of an old-fashioned lantern they&#39;re holding</pre><h3>Tips and Tricks for Maximizing Bing Video Creator</h3><p>To get the most out of your free Sora access, consider these advanced strategies:</p><h3>Optimizing Limited Duration</h3><p>The 5-second limitation can be challenging, but these approaches can help:</p><ol><li><strong>Focus on Single Moments</strong>: Design prompts around impactful single actions rather than complex sequences.</li><li><strong>Create Connected Series</strong>: Generate multiple related videos that can be edited together externally.</li><li><strong>Use Motion Efficiently</strong>: Request slow-motion effects for actions that benefit from detailed viewing.</li><li><strong>Loop-Friendly Design</strong>: Create scenes with natural loop points for platforms that auto-repeat content.</li></ol><h3>Managing Processing Time</h3><p>To work effectively with variable processing times:</p><ol><li><strong>Batch Creation</strong>: Submit multiple generation requests at once during off-peak hours.</li><li><strong>Staggered Workflow</strong>: Create prompts for tomorrow while today’s videos process.</li><li><strong>Strategic Rewards Usage</strong>: Save Fast generations for time-sensitive or urgent content needs.</li><li><strong>Monitor Processing Patterns</strong>: Note which types of prompts generate more quickly and use these patterns when time is limited.</li></ol><h3>Combining with Other Tools</h3><p>Extend the capabilities of your AI videos by using complementary tools:</p><ol><li><strong>Video Editing Software</strong>: Use tools like CapCut or Adobe Premiere to combine multiple AI clips.</li><li><strong>Audio Generation</strong>: Add AI-generated music or voiceovers using tools like Mubert or ElevenLabs.</li><li><strong>Text Overlays</strong>: Add captions, calls-to-action, or explanatory text through editing apps.</li><li><strong>Cross-Platform Publishing</strong>: Optimize your videos for different platforms using tools like Canva.</li></ol><h3>Building a Microsoft Rewards Strategy</h3><p>To maximize your Fast generation availability:</p><ol><li><strong>Daily Search Habit</strong>: Incorporate Bing searches into your daily routine to earn the maximum 150 points.</li><li><strong>Edge Browser Usage</strong>: Earn additional points by using Microsoft Edge for browsing.</li><li><strong>Complete Rewards Activities</strong>: Take advantage of bonus point opportunities in the Rewards dashboard.</li><li><strong>Strategic Point Usage</strong>: Save points for important projects requiring multiple Fast generations.</li></ol><h3>Future Developments and Alternatives in AI Video</h3><p>The AI video generation landscape is rapidly evolving, with several developments on the horizon:</p><h3>Planned Bing Video Creator Enhancements</h3><p>Microsoft has announced several upcoming improvements:</p><ul><li><strong>Desktop Access</strong>: Coming soon to Bing.com/create.</li><li><strong>Copilot Search Integration</strong>: Simplified access through Microsoft’s AI search experience.</li><li><strong>Horizontal Format Support</strong>: 16:9 aspect ratio option being added.</li><li><strong>Extended Duration</strong>: Potential increases to the 5-second limit based on user feedback.</li><li><strong>Enhanced Customization</strong>: More control over generated content elements.</li></ul><h3>Alternative Free AI Video Options</h3><p>While Bing Video Creator currently offers the most accessible Sora implementation, other options include:</p><ul><li><strong>Google’s Veo 3</strong>: Available through limited platforms but expanding access.</li><li><strong>Runway Gen-2</strong>: Offers a limited free tier with different model capabilities.</li><li><strong>Pika Labs</strong>: Discord-based video generation with free usage tiers.</li><li><strong>Stable Video Diffusion</strong>: An open-source alternative with various implementations.</li></ul><h3>The Evolving AI Video Ecosystem</h3><p>The introduction of free Sora access through Bing likely signals the beginning of broader democratization of AI video technology:</p><ul><li><strong>Increasing Competition</strong>: More platforms are likely to offer free tiers of advanced video AI.</li><li><strong>Feature Expansion</strong>: Capabilities like audio generation and longer durations are becoming standard.</li><li><strong>Integration with Existing Tools</strong>: Video creation platforms are incorporating AI generation features.</li><li><strong>Specialized Use Cases</strong>: Development of industry-specific video generation tools.</li></ul><h3>Conclusion: Harnessing the Power of Free Sora Access</h3><p>Microsoft’s integration of OpenAI’s Sora into Bing Video Creator marks a significant milestone in democratizing advanced AI video technology. While the free implementation has limitations compared to the paid version, it offers substantial value for content creators, educators, marketers, and curious users eager to explore AI-generated video without financial commitment.</p><p>By understanding the platform’s capabilities, crafting effective prompts, and implementing strategic workflows, users can leverage this free access to create impressive short-form content for social media, education, marketing, and creative projects. As the platform evolves and expands to desktop access, we can anticipate enhanced features and capabilities that will further bridge the gap between free and premium versions.</p><p>Whether you’re creating engaging social media content, visualizing complex educational concepts, or developing marketing materials, Bing Video Creator provides a compelling entry point into the world of AI-generated video. The democratization of this technology broadens creative possibilities for a much wider audience, allowing more people to benefit from the latest advancements in artificial intelligence.</p><blockquote><strong><em>Looking to integrate AI video capabilities into your applications?</em></strong><em> Apidog provides comprehensive API development, testing, and documentation tools to help you connect with services like OpenAI’s Sora. </em><a href="https://bit.ly/3Teeyxv"><em>Explore Apidog’s complete API lifecycle platform.</em></a></blockquote><h3>FAQ: Common Questions About Free Sora Access</h3><p><strong>Q: Do I need to pay anything to use Bing Video Creator with Sora?</strong> A: No, the basic service is completely free with a Microsoft account. The only optional cost is using Microsoft Rewards points for additional Fast generations beyond the initial 10.</p><p><strong>Q: How does the video quality compare between free and paid versions?</strong> A: Both use the same underlying Sora model, so visual quality is comparable, but the paid version offers higher resolution options and longer durations.</p><p><strong>Q: Can I use AI-generated videos commercially?</strong> A: Yes, Microsoft allows commercial use of content created with Bing Video Creator, similar to their policy for Bing Image Creator.</p><p><strong>Q: What happens to my videos after creation?</strong> A: Videos are stored on Microsoft’s servers for 90 days, during which you can download them to your device or share them through various channels.</p><p><strong>Q: Will the free version eventually support longer videos?</strong> A: Microsoft has indicated that feature expansions are planned, though specific timelines for extended duration support haven’t been announced.</p><p><strong>Q: Can I generate videos with specific people or copyrighted characters?</strong> A: No, both free and paid versions have content policies prohibiting the creation of videos featuring real identifiable people or copyrighted characters without authorization.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4ad60175796c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Unlocking On-Device AI: A Guide to Running Gemma 3n on Your Android Device]]></title>
            <link>https://medium.com/@Erik_Milosevic/unlocking-on-device-ai-a-guide-to-running-gemma-3n-on-your-android-device-c0f381340678?source=rss-f4e8d1747a77------2</link>
            <guid isPermaLink="false">https://medium.com/p/c0f381340678</guid>
            <category><![CDATA[android]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[gemma3n]]></category>
            <category><![CDATA[mobile-ai]]></category>
            <category><![CDATA[mobile-ai-tool]]></category>
            <dc:creator><![CDATA[Erik Milošević]]></dc:creator>
            <pubDate>Wed, 04 Jun 2025 09:29:17 GMT</pubDate>
            <atom:updated>2025-06-04T09:29:17.313Z</atom:updated>
            <content:encoded><![CDATA[<p>Mobile AI is transforming how we interact with technology, and Google’s Gemma 3n model is at the forefront of this revolution. Paired with the AI Edge Gallery, it offers a robust framework for deploying sophisticated language models directly onto Android devices. This guide will navigate you through setting up Gemma 3n on Android, leveraging Google’s cutting-edge tools for on-device inference.</p><blockquote><strong>Pro Tip:</strong> Supercharge your AI development! While Gemma 3n brings powerful on-device AI, managing the APIs that fuel your applications is crucial. <a href="https://bit.ly/3Teeyxv">Apidog</a> is your all-in-one API development platform, streamlining design, documentation, debugging, testing, and even mock server creation. It’s the perfect companion for building robust AI-integrated Android apps.</blockquote><h3>Gemma 3n &amp; Google AI Edge Gallery: A Powerful Duo</h3><p><a href="https://developers.googleblog.com/en/introducing-gemma-3n/">Gemma 3n</a> is Google’s answer to efficient, on-device language processing. Unlike cloud-dependent models, Gemma 3n runs locally, offering speed and privacy advantages.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*BvfUnrwkxBgm1IKp.png" /></figure><p>The Google AI Edge Gallery acts as a hub for AI deployment on edge devices. It’s packed with tools, samples, and documentation, simplifying the process of running models like Gemma 3n in resource-limited environments.</p><h3>Experience On-Device AI with Google AI Edge Gallery</h3><p><a href="https://github.com/google-ai-edge/gallery?tab=readme-ov-file">The Google AI Edge Gallery</a> is an experimental application that brings advanced Generative AI models to your Android device, operating entirely offline. It’s both a showcase and a sandbox for local AI model testing.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Cq8AVVWQYLq1rt91.png" /></figure><p>At its core, the Edge Gallery features optimized inference engines for model loading, memory management, and execution. A user-friendly interface allows interaction with models via text chat, image analysis, and multimodal inputs.</p><h3>Getting Started: Prerequisites</h3><p>To run Gemma 3n via the AI Edge Gallery, your Android device needs to meet certain specs: Android 8.0 (API level 26) or newer, at least 4GB RAM, and about 2GB of free storage for model files.</p><p>ARM64 architecture processors are recommended for best performance, though older ARM architectures are supported. Devices with NPUs or GPUs will see faster inference.</p><h3>Installation Walkthrough</h3><p>Since the AI Edge Gallery isn’t on the Google Play Store, you’ll install it manually via an APK from <a href="https://github.com/google-ai-edge/gallery?tab=readme-ov-file">GitHub</a>. Head to the releases section for the latest version.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*st9tiHEZ8JN0Wjby.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*bFKcQBBk-vPX6fWc.png" /></figure><p>First, enable “Install from Unknown Sources” in your Android settings (Settings &gt; Security &gt; Unknown Sources). Newer Android versions might prompt this per app during installation.</p><p>Download the latest APK (usually 50–100MB). Transfer it to your Android device (USB, cloud storage, or direct download).</p><p>Open the APK using a file manager to start the installation. Grant necessary permissions (storage, network access) when prompted.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ZzppdxRWMp1E5Mps.png" /></figure><p>Once installed, launch the AI Edge Gallery. Initial setup might take a few minutes as it configures the environment and downloads model components.</p><h3>Setting Up Gemma 3n Models</h3><p>With the AI Edge Gallery running, it’s time to download and configure Gemma 3n models. The app offers an easy way to manage models. You’ll need to download .task files from Hugging Face to get pre-configured Gemma 3n models optimized for mobile.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/588/0*81xt1AeyNzYXasPl.png" /></figure><p>Choose your model variant based on device capabilities and your needs. Smaller models are faster and use less memory but might be less capable. Larger models offer better performance but demand more resources.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/737/0*UCHtv69gIp9TsefH.png" /></figure><p>The app will show download progress and estimated completion times.</p><h3>Testing Your Setup</h3><p>Validate your Gemma 3n installation using the AI Edge Gallery’s built-in testing interfaces.</p><p>Start with text-based chats to check basic functionality. Responses should be quick (1–5 seconds). Keep an eye on system resource usage.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*N7Qwi67CQAQE3umc.png" /></figure><p>Next, explore multimodal features like image analysis. The app supports Ask Image (image-to-text), Prompt Lab (single-turn tasks), and AI Chat (multi-turn conversations), showcasing the platform’s versatility.</p><h3>Optimizing for Production</h3><p>Efficiently running Gemma 3n on Android involves several optimizations:</p><ul><li><strong>Memory Management:</strong> This is key. Use smart model loading and consider quantization to reduce memory footprint and boost inference speed without significant accuracy loss.</li><li><strong>Inference Scheduling:</strong> Minimize conflicts with other system processes using priority-based execution queues for responsive UI.</li><li><strong>Thermal Management:</strong> Prevent overheating during long AI sessions by monitoring device temperatures and implementing throttling if needed.</li></ul><h3>Integrating with Your Development Workflow</h3><p>Incorporating Gemma 3n into your Android projects requires robust tools. API testing is vital for ensuring smooth communication between AI components and your app’s logic.</p><p><a href="https://bit.ly/3Teeyxv">Apidog</a> is invaluable here. Its comprehensive testing suite helps validate API endpoints, response formats, and error handling — common challenges in AI apps.</p><p><a href="https://bit.ly/4e0MUfo">Sign Up for Free</a></p><p><a href="https://bit.ly/3XviWuE">Download Now</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Hr6DQEiwrxulNNHh.png" /></figure><p>When your app combines local AI with cloud services, Apidog’s mock server capabilities are a lifesaver for simulating various service conditions and testing resilience.</p><h3>What’s Next for Gemma 3n and AI Edge Gallery?</h3><p>The on-device AI landscape is evolving fast. Expect enhancements like better model compression, expanded multimodal capabilities (text, image, audio, video), and improved support for custom model fine-tuning. Google has also indicated that iOS support is on the horizon, broadening the reach of these powerful tools.</p><p>These advancements will empower developers to build even more specialized and sophisticated AI applications.</p><h3>Conclusion: Embracing On-Device AI</h3><p>Running Gemma 3n on Android via the Google AI Edge Gallery marks a significant leap in mobile AI. It empowers developers to create intelligent applications that run entirely on the user’s device, enhancing privacy and performance.</p><p>By following this guide’s technical advice on system requirements, installation, testing, and optimization, you can confidently deploy Gemma 3n and unlock new possibilities in mobile AI development.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c0f381340678" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Mastering API Documentation and Testing with Scalar]]></title>
            <link>https://medium.com/@Erik_Milosevic/mastering-api-documentation-and-testing-with-scalar-bb7c4ef099a6?source=rss-f4e8d1747a77------2</link>
            <guid isPermaLink="false">https://medium.com/p/bb7c4ef099a6</guid>
            <category><![CDATA[api-testing-tools]]></category>
            <category><![CDATA[scalar]]></category>
            <category><![CDATA[api-design]]></category>
            <category><![CDATA[api-documentation]]></category>
            <dc:creator><![CDATA[Erik Milošević]]></dc:creator>
            <pubDate>Thu, 22 May 2025 07:25:05 GMT</pubDate>
            <atom:updated>2025-05-22T07:25:05.293Z</atom:updated>
            <content:encoded><![CDATA[<p>Effective Application Programming Interfaces (APIs) are the backbone of modern software, but their value is significantly diminished without clear documentation and reliable testing. <a href="https://scalar.com/"><strong>Scalar</strong></a> emerges as a contemporary, open-source solution designed to address both needs, transforming standard OpenAPI specifications into interactive documentation and providing an integrated client for basic API testing. This guide explores how to leverage Scalar for creating professional API references and performing essential endpoint validation.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*IxrJPis_AfrS5lbl.png" /></figure><p><strong>Pro Tip:</strong> While Scalar offers excellent API documentation and basic testing, complex API workflows involving design collaboration, advanced automated testing, and robust mocking often require a more comprehensive solution. <a href="https://bit.ly/3Teeyxv"><strong>Apidog</strong> </a>integrates the entire API lifecycle into one platform, allowing teams to design, debug, test, and document APIs seamlessly. <a href="https://bit.ly/4e0MUfo">Elevate your API development process beyond documentation by exploring the powerful, integrated features at Apidog!</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/913/0*zcO_FQnQ1L1uM175.png" /></figure><h3>Understanding Scalar’s Core Offerings</h3><p>Scalar aims to modernize the often-static nature of API documentation. Its key features include:</p><ol><li><strong>Interactive Documentation:</strong> Renders OpenAPI (v3.0, v3.1) specifications into user-friendly, navigable web interfaces.</li><li><strong>Built-in API Client:</strong> Allows users to make live API calls directly from the documentation pages, facilitating exploration and testing.</li><li><strong>Code Sample Generation:</strong> Automatically generates request snippets in various languages (cURL, JavaScript, Python, etc.).</li><li><strong>Customization:</strong> Offers themes and CSS overrides to match branding.</li><li><strong>Flexible Deployment:</strong> Can be self-hosted, embedded into existing web projects via CDN, or used as a hosted service.</li></ol><p>By combining documentation and testing, Scalar streamlines the developer experience for both API producers and consumers.</p><h3>Initial Setup: Embedding Scalar via CDN</h3><p>For rapid prototyping or simple documentation needs, embedding Scalar using its CDN is the quickest way to get started.</p><p><strong>Steps:</strong></p><p><strong>1. Create an HTML File:</strong> Create a basic HTML file (e.g., api-docs.html).</p><p><strong>2. Add Scalar Boilerplate:</strong> Insert the following HTML structure, which loads the Scalar library and initializes it:</p><pre>&lt;!doctype html&gt;<br>&lt;html&gt;<br>&lt;head&gt;<br>  &lt;title&gt;My Scalar API Reference&lt;/title&gt;<br>  &lt;meta charset=&quot;utf-8&quot; /&gt;<br>  &lt;meta name=&quot;viewport&quot; content=&quot;width=device-width, initial-scale=1&quot; /&gt;<br>&lt;/head&gt;<br>&lt;body&gt;<br>  &lt;div id=&quot;app&quot;&gt;&lt;/div&gt;<br>  &lt;!-- Load Scalar --&gt;<br>  &lt;script src=&quot;https://cdn.jsdelivr.net/npm/@scalar/api-reference&quot;&gt;&lt;/script&gt;<br>  &lt;!-- Initialize Scalar --&gt;<br>  &lt;script&gt;<br>    Scalar.createApiReference(&#39;#app&#39;, {<br>      url: &#39;https://cdn.jsdelivr.net/npm/@scalar/galaxy/dist/latest.json&#39;,<br>      proxyUrl: &#39;https://proxy.scalar.com&#39;<br>    })<br>  &lt;/script&gt;<br>&lt;/body&gt;<br>&lt;/html&gt;&lt;!doctype html&gt; &lt;html&gt; &lt;head&gt;  &lt;title&gt;API Reference via Scalar&lt;/title&gt;  &lt;meta charset=&quot;utf-8&quot; /&gt;  &lt;meta name=&quot;viewport&quot; content=&quot;width=device-width, initial-scale=1&quot; /&gt; &lt;/head&gt; &lt;body&gt;  &lt;!-- Target container for Scalar --&gt;  &lt;div id=&quot;scalar-reference&quot;&gt;&lt;/div&gt;   &lt;!-- Load Scalar library --&gt;  &lt;script src=&quot;https://cdn.jsdelivr.net/npm/@scalar/api-reference&quot;&gt;&lt;/script&gt;   &lt;!-- Initialize Scalar --&gt;  &lt;script&gt;    Scalar.createApiReference(&#39;#scalar-reference&#39;, {      // Point to your OpenAPI specification URL (local or remote)      url: &#39;https://cdn.jsdelivr.net/npm/@scalar/galaxy/dist/latest.json&#39;, // Default example API      // Optional: Proxy URL for handling CORS in the built-in client      proxyUrl: &#39;https://proxy.scalar.com&#39;    });  &lt;/script&gt; &lt;/body&gt; &lt;/html&gt;</pre><p><strong>3. Open in Browser:</strong> Save the file and open it directly in your web browser. This will render the default Scalar Galaxy API reference.</p><p>This minimal setup requires no server or build process, making it ideal for initial exploration.</p><h3>Generating Documentation from an OpenAPI Specification</h3><p>Scalar dynamically generates documentation based on an OpenAPI file (JSON or YAML).</p><p><strong>Procedure:</strong></p><p><strong>1. Prepare Your OpenAPI File:</strong> Ensure you have a valid OpenAPI v3.x specification file. For example, a simple my-api.yaml:</p><pre>openapi: 3.0.3<br>info:<br>  title: My Sample API<br>  version: &#39;1.0&#39;<br>paths:<br>  /items:<br>    get:<br>      summary: Get all items<br>      description: Retrieve a list of available items.<br>      responses:<br>        &#39;200&#39;:<br>          description: Success<br>          content:<br>            application/json:<br>              schema:<br>                type: array<br>                items:<br>                  type: string</pre><p><strong>2. Make the Specification Accessible:</strong> Place the my-api.yaml file where your api-docs.html can access it. This could be:</p><ul><li>In the same directory (use a relative path: &#39;./my-api.yaml&#39;).</li><li>Hosted on a web server or CDN (use the full URL).</li><li><em>(Note: If loading locally without a server, browser security (CORS) might prevent loading. Running a simple local server like </em><em>python -m http.server is recommended.)</em></li></ul><p><strong>3. Update Scalar Configuration:</strong> Modify the url parameter in the Scalar.createApiReference function within your HTML file to point to your specification:</p><pre>Scalar.createApiReference(&#39;#scalar-reference&#39;, {<br>   url: &#39;./my-api.yaml&#39;, // Or your hosted URL<br>   proxyUrl: &#39;https://proxy.scalar.com&#39;<br> });</pre><p><strong>4. Reload:</strong> Refresh the api-docs.html page in your browser. Scalar will now parse your OpenAPI file and render the interactive documentation, including endpoints, descriptions, schemas, and code examples.</p><p><strong>Customization:</strong> Enhance the appearance using options within the configuration object:</p><pre>Scalar.createApiReference(&#39;#scalar-reference&#39;, {<br>  url: &#39;./my-api.yaml&#39;,<br>  proxyUrl: &#39;https://proxy.scalar.com&#39;,<br>  theme: &#39;dark&#39;, // Available themes: &#39;light&#39;, &#39;dark&#39;, &#39;moon&#39;, &#39;purple&#39;, &#39;solarized&#39;, &#39;kepler&#39;, etc.<br>  layout: &#39;modern&#39;, // Try &#39;classic&#39;<br>  showSidebar: true,<br>  // Add custom CSS rules if needed<br>  customCss: &#39;.scalar-api-reference { font-family: &quot;Arial&quot;, sans-serif; }&#39;<br>});</pre><h3>Performing API Tests with the Integrated Client</h3><p>Scalar’s built-in client allows for immediate testing of the documented endpoints.</p><p><strong>Workflow:</strong></p><ol><li><strong>Load a Testable API Specification:</strong> Ensure the url in your Scalar configuration points to the OpenAPI spec of a <em>running</em> API. You can use public test APIs like <a href="https://reqres.in/">ReqRes.in</a> (https://reqres.in/api/openapi.yaml) or your own locally hosted API.</li><li><strong>Navigate to an Endpoint:</strong> In the rendered Scalar documentation, select an endpoint (e.g., GET /api/users).</li><li><strong>Activate the Client:</strong> Click the “Try it” or similar button associated with the endpoint/method. This opens the request configuration panel.</li><li><strong>Configure the Request:</strong></li></ol><ul><li>Fill in path parameters, query parameters, or request bodies as required by the endpoint.</li><li>Add necessary headers (e.g., Authorization tokens, Content-Type).</li></ul><p><strong>5. Send the Request:</strong> Click the “Send” button.</p><p><strong>6. Analyze the Response:</strong> Scalar executes the API call (using the configured proxy if necessary to handle CORS) and displays the response details:</p><ul><li>Status Code (e.g., 200 OK, 404 Not Found)</li><li>Response Headers</li><li>Response Body (formatted if JSON/XML)</li><li>Request Duration</li></ul><p>This immediate feedback loop is invaluable for verifying endpoint behavior directly alongside the documentation.</p><h3>Advantages for Development Teams</h3><p>Integrating Scalar provides several benefits:</p><ul><li><strong>Single Source of Truth:</strong> Documentation is generated directly from the OpenAPI spec, reducing discrepancies.</li><li><strong>Improved Developer Experience:</strong> Interactive docs and built-in testing lower the barrier for API consumers to understand and integrate.</li><li><strong>Faster Iteration:</strong> Quickly test changes and validate endpoint behavior without switching to separate tools like Postman or cURL for basic checks.</li><li><strong>Modern Aesthetics:</strong> Presents APIs in a clean, visually appealing format compared to older documentation generators.</li></ul><h3>Conclusion</h3><p>Scalar offers a compelling blend of API documentation generation and basic testing capabilities within a modern, open-source framework. By leveraging OpenAPI specifications, it provides an efficient way to create interactive, user-friendly API references. The integrated testing client further enhances its utility, allowing for quick validation of endpoint functionality directly within the documentation context. Whether embedded in a project or used as a standalone tool, Scalar is a valuable asset for any developer or team working with APIs.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bb7c4ef099a6" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>