<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Mahesh</title>
    <description>The latest articles on DEV Community by Mahesh (@devnamipress).</description>
    <link>https://dev.to/devnamipress</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3936170%2F9dc99ef5-7101-455e-9002-b7dd23357831.jpeg</url>
      <title>DEV Community: Mahesh</title>
      <link>https://dev.to/devnamipress</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/devnamipress"/>
    <language>en</language>
    <item>
      <title>How to Create an AI Agent using Vercel Eve</title>
      <dc:creator>Mahesh</dc:creator>
      <pubDate>Sat, 04 Jul 2026 14:28:23 +0000</pubDate>
      <link>https://dev.to/devnamipress/how-to-create-an-ai-agent-using-vercel-eve-30nf</link>
      <guid>https://dev.to/devnamipress/how-to-create-an-ai-agent-using-vercel-eve-30nf</guid>
      <description>&lt;p&gt;Vercel has created an AI agent development framework to support building agents. This framework works similar to Google's agent development kit that has web module. &lt;/p&gt;

&lt;p&gt;So if you wish to create an agent that works on Vercel's stack, then Eve framework would be a good starting point. By default it has a decent UI too which makes running even simple web based agents easier. &lt;/p&gt;

&lt;p&gt;You can also combine Google ADK with Vercel Eve and build lot more powerful and autonomous agents. &lt;/p&gt;

&lt;p&gt;Take a look at the example below that I created using Vercel's Eve where I created AI agent with nutritional analysis from FDA. &lt;/p&gt;

&lt;p&gt;  &lt;iframe src="https://www.youtube.com/embed/S7LEoBdlBxA"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What does this AI Agent do?&lt;/strong&gt;&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;It has some pre-filled queries that answer your queries based on the respective queries. &lt;/li&gt;
&lt;li&gt;It can make fresh queries to USDA's API and fetch the health and calories specific information. &lt;/li&gt;
&lt;li&gt;You can also additionally use AI SDK and fetch different set of providers. &lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;In the process I also created an ADK router that covers most of the providers from the LiteLM's list. The &lt;a href="https://github.com/0xMaheshK/ADKRouter" rel="noopener noreferrer"&gt;repository ADKrouter&lt;/a&gt; has the options for both Python and typescript. It's Apache licensed so feel free to use it any way you please. &lt;/p&gt;

&lt;p&gt;Do note that there are plenty of OpenAI compatible API inference services. Take example of Openrouter which is a famous one. It can be used with the Free models from their offering to work with this agent. I did exactly that. &lt;/p&gt;

&lt;p&gt;You can check out my video above and let me know if you find that useful. &lt;/p&gt;

&lt;p&gt;How does the final process looks like? Agent has this sort of result I have produced. &lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fipvd07v9u42h4mq0oejw.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fipvd07v9u42h4mq0oejw.png" alt="ai-agent-eve-framework-devnami" width="800" height="822"&gt;&lt;/a&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>agentdevelopment</category>
      <category>tutorial</category>
      <category>automation</category>
    </item>
    <item>
      <title>How to Create Browser AI Agent using any LLM</title>
      <dc:creator>Mahesh</dc:creator>
      <pubDate>Tue, 30 Jun 2026 14:11:48 +0000</pubDate>
      <link>https://dev.to/devnamipress/how-to-create-browser-ai-agent-using-any-llm-1211</link>
      <guid>https://dev.to/devnamipress/how-to-create-browser-ai-agent-using-any-llm-1211</guid>
      <description>&lt;p&gt;I recently managed to get my hands on Browserbase. The platform that allows you to start a session with browser and get to extract data, manipulate browser for certain actions and build apps based on that data. &lt;/p&gt;

&lt;p&gt;Here's what you need to build a browser AI agent. &lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Google Antigravity IDE&lt;/li&gt;
&lt;li&gt;Browserbase account&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;You can use any agentic coding tools so it's not something you're limited to. Its just my preference. You can use other browser APIs too so don't limit yourself to a provider. &lt;/p&gt;

&lt;p&gt;I managed to give the whole process in this video below which you can check out step by step. &lt;/p&gt;

&lt;p&gt;  &lt;iframe src="https://www.youtube.com/embed/zvBYKTDCUlw"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;

&lt;p&gt;First thing is make sure to sign up to the Browserbase platform. And save the API key that you can get from the dashboard. &lt;/p&gt;

&lt;p&gt;Second thing you can do is open agentic coding tool. I am using Antigravity IDE as it is easy to use like cursor. And it has Gemini 3.5+ onwards models to work with. &lt;/p&gt;

&lt;p&gt;Third thing is you need a prompt. You have to explain your goal to the tool.&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;"Create a single page app that would scrape the Google search results for certain business. And extract data and displays it in an accessible yet minimal and presentable ways.Use the browserbase API key (attach). And here are the docs to browserbase: ref1, ref2"&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;From here onwards Geminin would give you an implementation plan. And you also get the walkthrough once the app is generated. Check if it does what it is intended to do. Also this is just one shot implementation. So at the end you have to refine it for final touch. &lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fdaqpxoq40065ezi5450b.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fdaqpxoq40065ezi5450b.png" alt="create-browser-ai-agent" width="800" height="374"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Above image is the final result of my prompt. And it is pretty decent app which i restricted to a certain business type. &lt;/p&gt;

&lt;p&gt;Like this you can create many browser agents that serve a specific purpose. &lt;/p&gt;

</description>
      <category>ai</category>
      <category>webscraping</category>
      <category>automation</category>
      <category>playwright</category>
    </item>
    <item>
      <title>Hermes Agent Desktop App for Windows | Free Local AI</title>
      <dc:creator>Mahesh</dc:creator>
      <pubDate>Fri, 05 Jun 2026 13:30:22 +0000</pubDate>
      <link>https://dev.to/devnamipress/hermes-agent-desktop-app-for-windows-free-local-ai-487j</link>
      <guid>https://dev.to/devnamipress/hermes-agent-desktop-app-for-windows-free-local-ai-487j</guid>
      <description>&lt;p&gt;I started with Hermes agent for some time now. I also recently wrote about my journey with setting up Hermes agent on the Linux OS. And turns out it is going on good so far. &lt;/p&gt;

&lt;p&gt;I learned how to set up the Hermes agent on WSL, Ubuntu Linux. So that's two different experience I have so far. I have yet to get it on Mac and do anything in automation there. &lt;/p&gt;

&lt;p&gt;Recently I learned about Hermes agent has released the desktop. And they are competing with the Google Spark, Antigravity and Codex on desktop app front. &lt;/p&gt;

&lt;p&gt;I covered the setup part here in the video if you are interested. &lt;/p&gt;

&lt;p&gt;  &lt;iframe src="https://www.youtube.com/embed/cw9DrXebl18"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;

&lt;p&gt;Now few days of experience so far taught me a lot about this desktop app. So I have some complaints and some good pat on the back too. &lt;/p&gt;

&lt;p&gt;It's not native build. Considering that part they didn't even bothered about asking user consent on which directory to install. That's the sad part, it goes to C: drive and consumes space. &lt;/p&gt;

&lt;p&gt;Default directory goes into C drive and making it fragmented and scattered. Fortunately you get to change this workspace directory. &lt;/p&gt;

&lt;p&gt;Openrouter is pain in the app. You may come across lack of tagging for the free models. And so if you have no credits you may end up facing 404 errors on the app terminal. &lt;/p&gt;

&lt;p&gt;Nvidia NIM Build also acts up. Like some of the models are extreme traffic on them. And so responses take forever. I don't blame the Hermes agent for this but atleast they should make it easier for us to choose our own custom APIs and we can fetch our providers there. Trying to be everything for everyone by spending resources on giving provider support often breaks the rest. &lt;/p&gt;

&lt;p&gt;Rest you can connect to other free providers and begin the chat or the coding. I have decided not to move with it as I am kind of happy with the Cursor, Antigravity and the Devin setup so far. &lt;/p&gt;

&lt;p&gt;If you are looking for UI version of Hermes agent then this is a good adventure to try out. But eventually coder in you would want to settle with the CLI. &lt;/p&gt;

</description>
      <category>ai</category>
      <category>hermesagent</category>
      <category>hermesagentdesktop</category>
    </item>
    <item>
      <title>Building Health Record App with Gemma 4</title>
      <dc:creator>Mahesh</dc:creator>
      <pubDate>Sat, 23 May 2026 05:22:35 +0000</pubDate>
      <link>https://dev.to/devnamipress/building-health-record-app-with-gemma-4-3d10</link>
      <guid>https://dev.to/devnamipress/building-health-record-app-with-gemma-4-3d10</guid>
      <description>&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/google-gemma-2026-05-06"&gt;Gemma 4 Challenge: Write About Gemma 4&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;When I discovered the agentic coding I ran out of ideas. Yes despite being a coder and IT experience. I just found myself frozen and can't think of solving any nearest problems at first. &lt;/p&gt;

&lt;p&gt;That was upto the point where I saw other people building vibe coding applications one after another. And that triggered me to think of my own problems. &lt;/p&gt;

&lt;p&gt;One of the problems I have is keeping track of health expenses and keeping track of health issues. I know there are many apps out there that lets you do that. But none of them were personal. &lt;/p&gt;

&lt;p&gt;So I came up with a prompt. &lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Create an app that let's you record health issues, medicine expenses, doctor visits, health issues and symptoms, medicine consumption streak and insurance data. Use Vite framework and React, SQlite.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;Sounds like very generic isn't it? No. If you check I was focused on making use of SQlite to keep my data local. And also wanted to use react to cut down the time. &lt;/p&gt;

&lt;p&gt;I used Kilo code and hooked into the Gemma 4. Which you can check in the progress of this app through the video below for the final results. &lt;/p&gt;

&lt;p&gt;  &lt;iframe src="https://www.youtube.com/embed/8AKQxl0lVgI"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;

&lt;p&gt;First step was turning single prompt into an app. Second adding tests to make sure each entry into database and the retrieval for results works. Third making sure the app is usable locally. &lt;/p&gt;

&lt;p&gt;So why not hosting on web? As this was personal project so I specifically decided to stay with the SQlite. And so keeping it offline only app. I could use watermelon or similar other database that would help it keep MVP on web. &lt;/p&gt;

&lt;p&gt;The final result looks like this. &lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fauc75hm29k68cliv53x5.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fauc75hm29k68cliv53x5.png" alt="Health App with Gemma4" width="800" height="541"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;So where to from here? I can think of creating mobile app or maybe think of making this available through Vercel or netlify hosting. And maybe even opensource this if I could not maintain this myself for long term. &lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>gemmachallenge</category>
      <category>gemma</category>
    </item>
    <item>
      <title>How to Setup Hermes Agent on Ubuntu Linux</title>
      <dc:creator>Mahesh</dc:creator>
      <pubDate>Thu, 21 May 2026 16:41:49 +0000</pubDate>
      <link>https://dev.to/devnamipress/how-to-setup-hermes-agent-on-ubuntu-linux-40lf</link>
      <guid>https://dev.to/devnamipress/how-to-setup-hermes-agent-on-ubuntu-linux-40lf</guid>
      <description>&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hermes-agent-2026-05-15"&gt;Hermes Agent Challenge&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Hermes agent has managed to catch up attention of automation engineers and the developers in just 3 months of establishing itself. Almost everyone around me are trying to setup Hermes on their VPS or the mobile phone by setting up a Linux VM. &lt;/p&gt;

&lt;p&gt;Earlier I thought it would require a VPS setup in order to do most of the integration like Openclaw has. Then I came across some of the YouTubers who setup Hermes agent and managed to everything including agentic chat for creating tasks and scheduled jobs to get the work done. &lt;/p&gt;

&lt;p&gt;I couldn't just stand watching other people set things up while I just get to play it safe not getting my hands dirty. So I managed to setup the Hermes Agent on Ubuntu Linux. &lt;/p&gt;

&lt;p&gt;Below I have documented the visual journey on How to Setup Hermes agent on Ubuntu. It was installed on Ubuntu 26.04 LTS version. &lt;/p&gt;

&lt;p&gt;  &lt;iframe src="https://www.youtube.com/embed/qIVcSuzvPMk"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;

&lt;p&gt;Here's what you should know. &lt;/p&gt;

&lt;p&gt;Basically you run the following command and follow the steps. You have option to follow recommended setup or manually setup the required steps. &lt;/p&gt;

&lt;p&gt;&lt;code&gt;curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash&lt;/code&gt;&lt;/p&gt;

&lt;p&gt;If you are new to this, best option would be go with some defaults. That would get your to the setup onboarding lot faster.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What should be the minimum hardware?&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;You need minimal 4GB onwards RAM on your computer. I had it on the 8GB Ubuntu VM and turns out it works smooth. &lt;/p&gt;

&lt;p&gt;You need an API key with Openrouter or Ollama cloud. With which you can use the free models from their catalog and still get to use the Hermes agent like premium user.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Which models are good to try out?&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Minimax M2.5&lt;/li&gt;
&lt;li&gt;Kimi K2.5&lt;/li&gt;
&lt;li&gt;Nvidia Nemotron &lt;/li&gt;
&lt;li&gt;gpt-oss&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;These are often the free and low cost models you can find in most of the API services out there. &lt;/p&gt;

&lt;p&gt;You can integrate faster if you use the web UI or the desktop shell for the Hermes agent. But that's not required you can still integrate from command line. But it becomes easier and faster if you use desktop or any Visual setup. &lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What you should do once you get comfortable with desktop setup of Hermes agent?&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;If you can afford move to VPS, some box as small as 4-6GB of RAM and 60GB SSD would be a good starting point to use the Hermes agent from the cloud. Integrate all the required tools and you would get more powerful setup to access from phone and web. &lt;/p&gt;

&lt;p&gt;That's it. Train your Hermes agent and enjoy your Automations. &lt;/p&gt;

</description>
      <category>hermesagentchallenge</category>
      <category>devchallenge</category>
      <category>agents</category>
      <category>hermesagent</category>
    </item>
  </channel>
</rss>
