DEV Community: Vivek Shetye

🚀 Build a Fully Local AI Agent with Hermes Agent, Ollama, Qwen 3.5, and SearXNG (100% Private & $0 Cost)

Vivek Shetye — Mon, 08 Jun 2026 20:02:58 +0000

What if you could build an AI agent that can:

✅ Think and reason

✅ Search the web

✅ Read and write files

✅ Generate reports and dashboards

✅ Run entirely on your own machine

Without:

❌ OpenAI API keys

❌ Anthropic subscriptions

❌ Monthly AI bills

❌ Sending your prompts and files to third-party servers

That's exactly what I built.

In this tutorial, I'll show you how to create a fully local AI agent stack using:

🤖 Hermes Agent

🧠 Qwen 3.5 9B via Ollama

🔎 SearXNG

The result is a powerful AI agent that costs $0 to operate, keeps your data private, and gives you complete control over your AI infrastructure.

🎥Full video walkthrough:

🤔 Why Build a Local AI Agent?

Most AI agents today depend on cloud APIs.

Every prompt, file, and conversation gets sent to someone else's servers.

For many use cases, that's perfectly fine.

But what if you're working with:

🔒 Sensitive business information

🔒 Private research data

🔒 Customer documents

🔒 Internal company knowledge

🔒 Personal notes and files

In those scenarios, privacy matters.

A local AI agent means:

✅ Your data never leaves your machine

✅ No third-party access to your prompts

✅ No API costs

✅ No rate limits

✅ Full ownership of your stack

And thanks to modern open-source models, local AI is becoming surprisingly capable.

🏗️ The Architecture

Our stack consists of three components.

🤖 Hermes Agent

Hermes Agent is an open-source AI agent framework developed by Nous Research.

Instead of just chatting with an LLM, Hermes turns the model into a true agent with:

Memory
Tool usage
Workflows
File access
Web search
Task execution

Think of it as the operating system for your AI agent.

🧠 Qwen 3.5 9B via Ollama

Next comes the brain.

We're using Qwen 3.5 9B running locally through Ollama.

Ollama makes it incredibly easy to run modern open-source language models on your machine.

The model handles:

Reasoning
Planning
Decision making
Report generation
Tool calling

And because it's running locally, every token stays on your hardware.

🔎 SearXNG

The final piece is SearXNG.

SearXNG is a privacy-focused meta search engine.

Instead of tracking users like traditional search providers, it aggregates results from multiple search sources while preserving privacy.

For AI agents, this means:

✅ Web search capabilities

✅ No tracking

✅ Self-hosted infrastructure

✅ Complete control

⚡ What Makes This Stack Interesting?

Most developers assume AI agents require expensive cloud infrastructure.

But with this setup:

💰 API Cost = $0

🔒 Data Privacy = 100%

⚙️ Infrastructure Ownership = 100%

🛠️ Customization = Unlimited

Everything runs locally.

Everything remains under your control.

🎯 Real Demo

To test the setup, I gave the agent a simple task:

Find the latest AI news and create an HTML report.

Here's what happened.

Step 1

The agent used SearXNG to search the web.

Step 2

It gathered and synthesized information from multiple sources.

Step 3

It generated a structured HTML report.

Step 4

The file was saved locally on my machine.

No cloud APIs.

No external AI providers.

No third-party processing.

Just a fully local AI agent doing real work.

🔥 The Best Part: It Scales

One thing I love about this architecture is that it grows with your hardware.

Starting point:

🧠 Qwen 3.5 9B

Future upgrades:

🚀 Larger Qwen models

🚀 70B parameter models

🚀 400B parameter models

🚀 Multi-GPU setups

The architecture stays exactly the same.

You simply swap in a more capable model.

The only real limitation is your hardware.

💡 Potential Use Cases

Developers are already building some fascinating things with local AI agents.

Examples include:

📚 Research assistants

📄 Private document analysis

💻 Coding assistants

📈 Market research workflows

📰 News aggregation systems

📋 Report generation pipelines

🏢 Internal company knowledge assistants

🔬 Scientific research agents

🔒 Privacy-first enterprise AI solutions

Because everything is self-hosted, these use cases become much easier to justify from a security and compliance perspective.

🌍 Why Local AI Is Becoming a Big Deal

The AI industry spent the last few years moving everything to the cloud.

Now we're seeing another trend emerge:

Bringing AI back to the device.

Open-source models are improving rapidly.

Consumer hardware is becoming more powerful.

Agent frameworks are becoming more capable.

As a result, local AI is no longer just a hobby project.

It's becoming a practical option for real-world applications.

The combination of:

🤖 AI Agents

🧠 Open Models

🔒 Privacy

💰 Zero API Cost

is incredibly compelling.

💬 What Would You Build?

If you had a fully private AI agent running entirely on your own machine...

What would you build?

A coding assistant?

A research agent?

A private knowledge system?

A business automation workflow?

Let me know in the comments. I'm always curious to see what developers are creating with local AI.

🚀 Hermes Agent Just Added a Kanban Board — And It Changes Everything

Vivek Shetye — Thu, 04 Jun 2026 18:59:26 +0000

Most AI agent frameworks have a problem.

You can create agents.

You can give them tools.

You can even make them collaborate.

But as soon as your workflow gets complex, things start breaking down.

📌 Tasks become difficult to track.

📌 Dependencies become messy.

📌 You lose visibility into what’s running.

📌 Outputs become disconnected.

📌 Coordination becomes manual.

The new Hermes Agent Kanban Board solves exactly that problem.

And after spending time with it, I genuinely think this is one of the biggest upgrades Hermes Agent has shipped so far.

To test it, I decided to orchestrate an entire product launch campaign using multiple specialized AI agents working together inside a single workflow.

What happened was pretty impressive.

🤔 Why Most AI Workflows Fall Apart

Most AI tools today are conversation-driven.

The workflow usually looks like this:

💬 Open a chat

✍️ Write a prompt

📄 Copy the output

📋 Paste it somewhere else

🔄 Repeat 20 more times

Eventually you’re juggling:

Multiple chat windows
Notion pages
Google Docs
Spreadsheets
Task trackers

The AI may be smart.

But the workflow isn’t.

And that’s where Hermes Kanban enters the picture.

🎯 What Exactly Is Hermes Kanban?

Hermes Kanban is a visual workflow orchestration system built directly into Hermes Agent.

Instead of managing agents through separate chats, you coordinate everything through a task board.

You get:

✅ Task assignment

✅ Dependencies

✅ Parent-child relationships

✅ Parallel execution

✅ Live logs

✅ Workflow visualization

✅ Artifact management

Think of it as:

🧠 AI Agents + 📋 Trello + 🎯 Workflow Automation

all inside a single system.

🎥 Watch The Full Walkthrough

In the video, I cover:

✅ Installing Hermes Agent

✅ Creating specialist agents

✅ Setting up web search

✅ Building Kanban workflows

✅ Managing dependencies

✅ Running tasks in parallel

✅ Reviewing outputs

✅ Publishing final deliverables

🔥 The Feature That Immediately Got My Attention

Task Dependencies.

This sounds simple.

But it’s incredibly powerful.

Imagine you’re launching a product.

Before creating content, you need:

🔍 Audience Research

Then:

📊 Messaging & Positioning

Only after that can you create:

✍️ Blog Posts

📧 Email Sequences

📱 Social Content

🎥 YouTube Scripts

With Hermes Kanban, those relationships are explicitly defined.

Audience Research
       ↓
Messaging & Positioning
       ↓

Landing Page
Blog Posts
Emails
Social Posts
YouTube Scripts

Every downstream task can reference outputs generated by upstream tasks.

This creates a true workflow rather than a collection of isolated prompts.

⚡ Parallel Agent Execution Is Where Things Get Interesting

Once positioning was complete, I launched five content-generation tasks simultaneously.

Suddenly I had:

✍️ Landing Page Creation

📝 Blog Writing

📧 Email Sequence Generation

📱 Social Media Content

🎥 YouTube Script Creation

all running at the same time.

Not one after another.

Not manually triggered.

The system orchestrated everything automatically.

Watching multiple AI agents execute work in parallel felt less like using an AI tool and more like managing a real team.

👥 Building a Team of Specialized AI Agents

For this demo, I created four specialist agents.

🔬 Researcher

Responsible for:

Competitor analysis
Audience discovery
Market research
Pain point analysis

⸻

📊 Analyst

Responsible for:

Positioning
Messaging
Strategic narratives
Market differentiation

⸻

✍️ Writer

Responsible for:

Landing pages
Blogs
Emails
Social content
Video scripts

⸻

🔍 Reviewer

Responsible for:

Quality control
Consistency checks
Accuracy validation
Brand alignment

Each agent had its own:

🧠 Memory

⚙️ Configuration

📖 Instructions

🎯 Responsibilities

This specialization dramatically improves output quality.

👀 Seeing What Your Agents Are Doing

One thing I really appreciated was transparency.

Every task provides:

📜 Execution logs

📊 Status updates

📂 Generated artifacts

⏱️ Progress tracking

Instead of wondering:

“What is my agent doing right now?”

You can actually see it.

For complex workflows, this is incredibly valuable.

🚀 My Demo Workflow: Launching Momentum

To showcase the Kanban board, I created a launch workflow for a fictional product called Momentum.

Momentum is an AI-powered habit tracker that adapts to your natural energy patterns.

The workflow looked like this:

🔍 Audience Research

↓

📊 Messaging & Positioning

↓

✍️ Landing Page

📝 Blog Posts

📧 Email Sequence

📱 Social Posts

🎥 YouTube Scripts

↓

🔍 Content Review

↓

✅ Final Revisions

The marketing campaign itself wasn’t the interesting part.

The interesting part was watching Hermes Kanban coordinate the entire process.

🧩 The Reviewer Agent Was Surprisingly Useful

Once all content was generated, I handed everything over to the Reviewer agent.

It analyzed:

📧 Emails

📄 Landing Pages

📝 Blog Content

📱 Social Posts

🎥 Scripts

and checked for:

✅ Consistency

✅ Clarity

✅ Accuracy

✅ Tone

It identified several issues and produced actionable feedback.

Then I created one final task.

The Writer agent consumed that feedback and automatically updated all assets.

This created a genuine feedback loop between agents.

Exactly how real teams operate.

🌎 This Goes Far Beyond Marketing

The marketing workflow was simply a demonstration.

The same approach could be used for:

💻 Software Development

Research → Design → Code → Review → Documentation

📈 SEO Operations

Keyword Research → Content Brief → Writing → Optimization

🧪 Research Projects

Data Collection → Analysis → Reporting

🚀 Startup Execution

Research → Strategy → Outreach → Growth

📰 Content Teams

Research → Writing → Editing → Publishing

Anywhere you have a repeatable workflow, Hermes Kanban becomes interesting.

💡 Why I Think Hermes Kanban Is A Big Deal

Most AI products focus on conversations.

Hermes Kanban focuses on workflows.

That’s a major difference.

Instead of:

👤 Human → 🤖 AI

you get:

👤 Human → 👥 AI Team

One agent researches.

One analyzes.

One writes.

One reviews.

The workflow coordinates everything.

You supervise.

That feels much closer to the future of AI systems than simply chatting with a chatbot.

💬 What Would You Build?

The most exciting part of Hermes Kanban isn’t marketing automation.

It’s the ability to coordinate teams of AI agents through structured workflows.

I’m curious:

👉 What workflow would you automate first?

Product development?

Content operations?

Research?

Customer support?

Let me know in the comments.

🚀 Build a Self-Improving AI Assistant with Hermes (Beginner-Friendly Step-by-Step Guide)

Vivek Shetye — Thu, 07 May 2026 19:26:38 +0000

Most AI tools today are still glorified chatbots.

You ask a question.
You get an answer.
And by tomorrow… it forgets you even exist.

But what if you could build an AI assistant that:

🧠 Remembers your preferences
📈 Learns from every interaction
🌐 Researches the web in real time
📱 Runs directly on your phone
⏰ Automates tasks while you sleep

That’s exactly what I built using Hermes, one of the fastest-rising AI agent frameworks right now and increasingly seen as a serious competitor in the personal AI space.

🔥 Why Hermes Is Blowing Up

Hermes isn’t just another chatbot framework.

It introduces something much more powerful:

🧠 Long-Term Memory

Your assistant remembers:

Your goals
Your recurring tasks
Your preferences
Your habits

🔄 Self-Reflection

After every task, Hermes can evaluate:

What worked
What failed
What should improve next time

This means your AI literally gets smarter the more you use it.

🤖 Real Automation

You can use Hermes for:

Daily AI news briefings
Travel planning
Productivity reminders
Research tasks
Personal scheduling
Coding assistant
Messaging integrations like Telegram,Discord, etc

🛠️ What You’ll Build in This Tutorial

In this beginner-friendly walkthrough, I cover:

✅ Installing Hermes from scratch
✅ Configuring free Google AI Studio API access
✅ Adding Tavily for live web search
✅ Connecting your AI assistant to Telegram
✅ Creating a true phone-based personal assistant
✅ Security best practices

🎥 Full video walkthrough

⚙️ Installation: Build Your First Hermes AI Assistant

Getting started with Hermes is surprisingly simple — even if you’re completely new to AI agents.

Within minutes, you’ll have your own self-improving assistant running locally.

🚀 Step 1: Run the Hermes Installer

Simply run:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

What this does:

✅ Downloads Hermes
✅ Installs required dependencies
✅ Configures your environment
✅ Guides you through setup interactively

💡Pro Tip: macOS and Linux users can install directly. Windows users should first install WSL2 for the smoothest experience.

🤖 Step 2: Choose Your Model Provider

Hermes supports multiple providers, but for beginners, Google AI Studio (Gemini) is one of the best choices because:

💸 Generous free tier
⚡ Fast setup
🛠️ Beginner-friendly
🚀 Great for experimentation

During setup, select:

(•) Google AI Studio (Gemini models — native Gemini API)

🔑 Step 3: Get Your Google AI Studio API Key

To power Hermes, you’ll need an API key.

Steps:

Go to Google AI Studio
Navigate to Projects
Click Create New Project
Go to API Keys
Select your project
Click Create API Key
Copy your key
Paste it into the Hermes installer when prompted

🔒 Security Tip:
Treat API keys like passwords — never share them publicly.

📱 Step 4: Skip Messaging Setup (For Now)

During initial installation, Hermes will ask about messaging integrations.

👉 Skip this step temporarily.

We’ll configure Telegram later to transform Hermes into a true mobile personal assistant.

💬 Step 5: Launch Hermes Chat

When prompted, enter: y

Hermes will launch its built-in TUI (Terminal User Interface).

You now have:

✅ Your first AI agent
✅ Local chat interface
✅ Long-term memory capabilities
✅ Self-improvement foundation

🎉 Congratulations — your personal AI assistant is officially alive.

🔍 Configure Tavily for Real-Time Web Search

Without web access, your AI assistant is limited.

Adding Tavily gives Hermes the ability to:

🌐 Search the internet
📰 Gather live information
📊 Perform research tasks
✈️ Plan trips using current data

🛠️ Tavily Setup Steps

Create a Tavily account

Visit Tavily and sign up.

Generate your API key

Create a new key from your dashboard.

Add it to Hermes

hermes config set TAVILY_API_KEY <YOUR_TAVILY_API_KEY>

🎯 Result:

Your Hermes assistant can now perform real-time web searches and research autonomously.

📲 Connect Hermes to Telegram (Your AI in Your Pocket)

Right now, Hermes only runs on your computer.

Connecting Telegram transforms it into a true personal assistant you can message from anywhere.

🤖 Step 1: Create a Telegram Bot

Instructions:

Open Telegram
Search for @BotFather
Type: /newbot
Choose a bot name
Choose a unique username ending in: _bot
Copy your bot token

🔑 You’ll need this token to connect Hermes.

Step 2: Connect Telegram to Hermes

Run the following command:

hermes setup gateway

Then:

✅ Select Telegram
✅ Paste your Bot Token
✅ Add your Telegram numeric user ID
✅ Set your account as the home channel
✅ Start the gateway service

📍 How to Find Your Telegram User ID

Search Telegram for: @userinfobot

It will instantly provide your numeric user ID.

🎉 Final Result

Your Hermes AI Assistant is now:

📱 Available on your phone
💬 Reachable through Telegram
🧠 Memory-enabled
🌐 Web-connected
📈 Self-improving

You’ve officially built a beginner-friendly AI agent that can evolve into a true personal productivity system.

🌍 Real-World Examples

Once set up, your AI assistant can:

✈️ Personalized Travel Planning

Ask it to plan a trip to Japan… then later France…

And it remembers:

Your budget
Your interests
Your travel style

📰 Daily Briefings

Example:

“Every morning at 8 AM, send me the top 3 AI news stories.”

Your assistant works while you sleep.

📅 Weekly Productivity Systems

Example:

“Every Friday at 4 PM, remind me to review my week.”

⚠️ Important Security Tips

Hermes is powerful — but with power comes responsibility.

Best Practices:

🔒 Avoid running on your primary machine
🖥️ Use a VM, remote server, or sandbox
🔑 Protect API keys like passwords
⚡ Be cautious with community-built skills

🧠 Why This Is Bigger Than Just One Tutorial

We’re watching AI evolve from:

Old Model:

Chatbot → Answer → Forget

New Model:

Assistant → Learn → Remember → Improve

This is a massive shift.

We’re entering the era of persistent personal AI.

💬 Final Thoughts

Hermes shows where AI is truly heading:

✨ Personalized
✨ Persistent
✨ Autonomous
✨ Self-improving

For beginners, this may be one of the most practical AI agent projects you can build right now.

🚀 What would you automate first with your own personal AI assistant?

🚀 Build a Self-Improving AI Assistant with Hermes (Beginner-Friendly Step-by-Step Guide)

Vivek Shetye — Thu, 07 May 2026 19:26:38 +0000

Most AI tools today are still glorified chatbots.

You ask a question.
You get an answer.
And by tomorrow… it forgets you even exist.

But what if you could build an AI assistant that:

🧠 Remembers your preferences
📈 Learns from every interaction
🌐 Researches the web in real time
📱 Runs directly on your phone
⏰ Automates tasks while you sleep

That’s exactly what I built using Hermes, one of the fastest-rising AI agent frameworks right now and increasingly seen as a serious competitor in the personal AI space.

🔥 Why Hermes Is Blowing Up

Hermes isn’t just another chatbot framework.

It introduces something much more powerful:

🧠 Long-Term Memory

Your assistant remembers:

Your goals
Your recurring tasks
Your preferences
Your habits

🔄 Self-Reflection

After every task, Hermes can evaluate:

What worked
What failed
What should improve next time

This means your AI literally gets smarter the more you use it.

🤖 Real Automation

You can use Hermes for:

Daily AI news briefings
Travel planning
Productivity reminders
Research tasks
Personal scheduling
Coding assistant
Messaging integrations like Telegram,Discord, etc

🛠️ What You’ll Build in This Tutorial

In this beginner-friendly walkthrough, I cover:

🎥 Full video walkthrough

⚙️ Installation: Build Your First Hermes AI Assistant

Getting started with Hermes is surprisingly simple — even if you’re completely new to AI agents.

Within minutes, you’ll have your own self-improving assistant running locally.

🚀 Step 1: Run the Hermes Installer

Simply run:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

What this does:

✅ Downloads Hermes
✅ Installs required dependencies
✅ Configures your environment
✅ Guides you through setup interactively

💡Pro Tip: macOS and Linux users can install directly. Windows users should first install WSL2 for the smoothest experience.

🤖 Step 2: Choose Your Model Provider

Hermes supports multiple providers, but for beginners, Google AI Studio (Gemini) is one of the best choices because:

💸 Generous free tier
⚡ Fast setup
🛠️ Beginner-friendly
🚀 Great for experimentation

During setup, select:

(•) Google AI Studio (Gemini models — native Gemini API)

🔑 Step 3: Get Your Google AI Studio API Key

To power Hermes, you’ll need an API key.

Steps:

Go to Google AI Studio
Navigate to Projects
Click Create New Project
Go to API Keys
Select your project
Click Create API Key
Copy your key
Paste it into the Hermes installer when prompted

🔒 Security Tip:
Treat API keys like passwords — never share them publicly.

📱 Step 4: Skip Messaging Setup (For Now)

During initial installation, Hermes will ask about messaging integrations.

👉 Skip this step temporarily.

We’ll configure Telegram later to transform Hermes into a true mobile personal assistant.

💬 Step 5: Launch Hermes Chat

When prompted, enter: y

Hermes will launch its built-in TUI (Terminal User Interface).

You now have:

✅ Your first AI agent
✅ Local chat interface
✅ Long-term memory capabilities
✅ Self-improvement foundation

🎉 Congratulations — your personal AI assistant is officially alive.

🔍 Configure Tavily for Real-Time Web Search

Without web access, your AI assistant is limited.

Adding Tavily gives Hermes the ability to:

🌐 Search the internet
📰 Gather live information
📊 Perform research tasks
✈️ Plan trips using current data

🛠️ Tavily Setup Steps

Create a Tavily account

Visit Tavily and sign up.

Generate your API key

Create a new key from your dashboard.

Add it to Hermes

hermes config set TAVILY_API_KEY <YOUR_TAVILY_API_KEY>

🎯 Result:

Your Hermes assistant can now perform real-time web searches and research autonomously.

📲 Connect Hermes to Telegram (Your AI in Your Pocket)

Right now, Hermes only runs on your computer.

Connecting Telegram transforms it into a true personal assistant you can message from anywhere.

🤖 Step 1: Create a Telegram Bot

Instructions:

Open Telegram
Search for @BotFather
Type: /newbot
Choose a bot name
Choose a unique username ending in: _bot
Copy your bot token

🔑 You’ll need this token to connect Hermes.

Step 2: Connect Telegram to Hermes

Run the following command:

hermes setup gateway

Then:

✅ Select Telegram
✅ Paste your Bot Token
✅ Add your Telegram numeric user ID
✅ Set your account as the home channel
✅ Start the gateway service

📍 How to Find Your Telegram User ID

Search Telegram for: @userinfobot

It will instantly provide your numeric user ID.

🎉 Final Result

Your Hermes AI Assistant is now:

📱 Available on your phone
💬 Reachable through Telegram
🧠 Memory-enabled
🌐 Web-connected
📈 Self-improving

You’ve officially built a beginner-friendly AI agent that can evolve into a true personal productivity system.

🌍 Real-World Examples

Once set up, your AI assistant can:

✈️ Personalized Travel Planning

Ask it to plan a trip to Japan… then later France…

And it remembers:

Your budget
Your interests
Your travel style

📰 Daily Briefings

Example:

“Every morning at 8 AM, send me the top 3 AI news stories.”

Your assistant works while you sleep.

📅 Weekly Productivity Systems

Example:

“Every Friday at 4 PM, remind me to review my week.”

⚠️ Important Security Tips

Hermes is powerful — but with power comes responsibility.

Best Practices:

🔒 Avoid running on your primary machine
🖥️ Use a VM, remote server, or sandbox
🔑 Protect API keys like passwords
⚡ Be cautious with community-built skills

🧠 Why This Is Bigger Than Just One Tutorial

We’re watching AI evolve from:

Old Model:

Chatbot → Answer → Forget

New Model:

Assistant → Learn → Remember → Improve

This is a massive shift.

We’re entering the era of persistent personal AI.

💬 Final Thoughts

Hermes shows where AI is truly heading:

✨ Personalized
✨ Persistent
✨ Autonomous
✨ Self-improving

For beginners, this may be one of the most practical AI agent projects you can build right now.

🚀 What would you automate first with your own personal AI assistant?

Google Agents CLI + Claude Code: Building Production-Style AI Agents in Under 30 Minutes

Vivek Shetye — Tue, 28 Apr 2026 18:53:19 +0000

Google released something that could significantly accelerate how developers build AI agents:

Google Agents CLI

Combined with:

Google ADK (Agent Development Kit)
Claude Code (or Gemini CLI / OpenCode)

it creates one of the fastest workflows currently available for building, testing, evaluating, and deploying multi-agent systems.

In this project, I built a full Multi-agent Customer Support team in under 30 minutes.

What I Built

A production-style customer support team powered by four specialized AI agents:

🎧 Concierge Agent

First point of contact
User intent classification
Request routing

📦 Logistician Agent

Order status
Shipping updates
Inventory checks

🎭 Stylist Agent

Product recommendations
Catalog discovery
Personalized suggestions

🛡️ Resolver Agent

Returns
Refunds
Human escalation for high-value disputes

Full Video Walkthrough

Core Stack

Google ADK

Google’s Python-native Agent Development Kit that provides:

Agent abstractions
Tool integration
Session handling
Multi-agent architecture patterns

Google Agents CLI

A workflow layer that enables:

Scaffold
Build
Validate
Deploy

Claude Code

Your implementation accelerator:

Writes code
Generates tests
Creates evals
Performs security audits
Assists deployment

Workflow

1. Scaffold the Foundation with Google Agents CLI

The process starts by using Google Agents CLI to rapidly initialize and scaffold the entire multi-agent project structure.

This includes:

Base architecture
Agent framework setup
Development workflow
Deployment pathways

Instead of manually creating boilerplate, the CLI provides a production-oriented foundation from day one.

2. Define the Multi-Agent System Through Natural Language

Next, Claude Code acts as the implementation engine.

By providing detailed system requirements in plain language, I specified:

Individual agent roles
Responsibilities for each specialist
Agent-to-agent communication patterns
Human-in-the-loop workflows
Session memory requirements
Mock data sources
Deployment targets

This transforms high-level business logic directly into executable architecture.

3. Rapid End-to-End System Generation

From those instructions, Claude Code + Agents CLI collaboratively generated:

System Design:

Full design specification
Agent hierarchy
Routing logic
Communication workflows

Development Assets:

Agent definitions
Tool integrations
Mock datasets
Core application code

Quality Assurance:

Unit tests
Integration tests
Evaluation suites
Security audit recommendations

4. Deployment

The system successfully:

Containerized the application
Pushed to Artifact Registry
Configured IAM
Deployed to Google Cloud Run
Created GitHub Actions CI/CD workflows

Which means every future code push can:

Test → Eval → Deploy automatically

This workflow creates a streamlined path from concept → validated production prototype in dramatically less time than traditional development workflows.

Key Takeaway

The hardest part is no longer building AI agents.

It’s deciding what to build.

That’s a massive shift.

As tooling matures, developer leverage increases dramatically.

Production Advice

If you’re planning to use this stack seriously:

Prioritize:

Prompt injection defenses
Adversarial evals
Human oversight
Security hardening
Guardrails
Monitoring

Fast building does NOT remove production responsibility.

Final Thoughts

Google Agents CLI + Claude Code feels like an early glimpse into the future of AI product development.

For:

AI engineers
Startup founders
Automation builders
Developer tool creators

This workflow could meaningfully compress idea-to-production timelines.

Full Code Repository

👉 https://github.com/vivekshetye/google-adk-multi-agent-customer-support

🚀 I Built a Fully Autonomous AI Marketing Team (That Never Sleeps)

Vivek Shetye — Wed, 22 Apr 2026 15:11:21 +0000

What I Built

Marketing today isn’t just about creating content, it’s about research, strategy, distribution, and consistency. Doing all of that manually is slow, expensive, and honestly… hard to scale.

So I built something different.

👉 A fully autonomous AI Marketing Team powered by OpenClaw.

This system is made up of 4 specialized AI agents:

👑 Orchestrator Agent → The brain that plans and assigns tasks
🔎 TrendScout Agent → The researcher that finds real-world trends
📈 Growth Agent → SEO + GEO optimizer for discoverability
✍️ Copywriter Agent → Converts insights into high-quality content

All of these agents:

Live inside Discord
Communicate with each other
Maintain their own memory
Collaborate autonomously
Keep working until the job is done

💡 No micromanagement. No context switching. Just results.

To test this system, I created a sample SaaS product:
SpotSeeker — a platform for digital nomads to find verified workspaces.

And then I gave my AI team a simple task:

“Create a 30-day marketing launch plan for SpotSeeker in New York.”

What happened next was wild.

How I Used OpenClaw

OpenClaw is what made this entire system possible. It acts as the execution layer for multi-agent collaboration.

Here’s how I wired everything together:

🧠 1. Multi-Agent Architecture

Each agent runs as an independent entity inside OpenClaw, with:

Its own persona
Defined responsibilities
Separate memory (short-term + long-term)
Ability to communicate via Discord mentions

The key idea:

Instead of one “smart” agent, build a team of focused specialists.

🔗 2. Discord as the Communication Layer

I used Discord bots for each agent and connected them via OpenClaw.

Key setup:

Each agent mapped to a unique bot
Messages routed via bindings in config
Controlled access using channel allowlists
Only respond when mentioned
Enabled bot-to-bot communication
Instructed each agent on how to mention other agents

This setup ensures:
✔ Messages go to the right agent
✔ Agents don’t interrupt each other randomly
✔ Conversations stay structured
✔ True task handoff between agents.

🧵 3. Thread-Based Context Management

Instead of one noisy channel, I used Discord threads for each task.

Why this matters:

Keeps context clean
Prevents token bloat
Improves response quality

🧩 4. Skills + Tooling

Both the Research Agent and Growth Agent dynamically created a skill for:

Fetching Google Trends data
Handling rate limits (429 errors)
Pulling insights from the web (via SearXNG)

This allowed them to:

Identify trending cities
Extract keyword demand
Build SEO strategies

🤖 5. Model Choice: Minimax M2.7

I used Minimax M2.7, which performs extremely well for:

Multi-step reasoning
Agent coordination
Long workflows

Its strong agentic performance made the system feel surprisingly… reliable.

Demo

🎥 Full video walkthrough

🎥 Watch how a single prompt turns into a full marketing campaign

Orchestrator breaks the task into phases
Research Agent analyzes the NYC market
Growth Agent builds SEO + GEO strategy
Copywriter generates content
Agents collaborate, fix errors, and finalize output

📊 Final Output (generated in ~10 minutes):

15 LinkedIn posts
20 Twitter threads
3 YouTube video scripts

All:

Data-backed
SEO optimized
On-brand

👉 You can check the full setup, prompts, and configs on GitHub (linked in the video description).

What I Learned

1. Multi-Agent > Single Agent

A single LLM can do many things…

But a team of agents with clear roles performs way better.

It’s like hiring specialists instead of expecting one person to do everything.

2. Memory Changes Everything

These agents don’t just execute tasks — they remember context.

Think of it like hiring someone:

First task → rough
Feedback → improvement
Over time → they get better

That’s exactly what happens here.

3. Autonomy Needs Guardrails

Without:

Clear instructions
Output formats
Defined responsibilities

Agents can drift or loop.

The key is:

Give freedom, but with structure.

4. This Changes How We Build Teams

This isn’t just a demo.

It’s a glimpse into a future where:

Teams are hybrid (humans + agents)
Workflows are autonomous
Execution is near-instant

ClawCon Michigan

I didn’t attend ClawCon Michigan this time, but seeing what’s possible with OpenClaw definitely makes it an event I’d want to be part of in the future.

Final Thoughts

What surprised me the most wasn’t that this worked…

It’s how well it worked.

From a single prompt → to a complete marketing campaign
From zero → to execution in minutes

And the best part?

This system runs 24/7.

No burnout. No delays. No excuses.

🚀 I Built a Fully Local AI Agent for $0 (No Cloud, No API Costs)

Vivek Shetye — Wed, 15 Apr 2026 13:18:51 +0000

Everyone is talking about AI agents right now.

But most tutorials fall into one of two categories:

💸 You’re expected to spend $100–$200/month on APIs/LLM subscriptions
🖥️ Or you need a powerful GPU setup to run local models

I wanted something different.

👉 Could I build a fully working AI agent that runs locally on just a laptop — for $0?

So I tried.

And what I ended up building was more powerful than I expected.

🧠 What I Built

I built a proactive AI agent that runs entirely on my laptop inside a VM.

It can:
• 💬 Talk to me on Telegram
• 🔎 Search the web privately
• 📁 Write and manage files
• 🧠 Maintain memory across conversations
• 🤖 Execute multi-step research tasks

And the best part?

💰 Total cost: $0

No subscriptions. No API bills. No cloud infrastructure.

⚙️ The Stack Behind It

This system is built using three core components:

🧩 OpenClaw — The Agent Framework

Think of this as the nervous system.

It handles:
• Tool usage
• Memory management
• Decision-making flow
• Message routing between components

⚠️ It’s still in beta, so expect some rough edges, but the architecture is powerful.

⚡ Gemini 3.1 Flash Lite — The Brain

This powers the reasoning layer.

Free tier includes:
• 15 requests/min
• 500 requests/day
• 250K tokens/min

Perfect for:
• Learning agent workflows
• Multi-step tasks
• Rapid experimentation

It’s surprisingly fast, which matters a lot in agent loops.

🔍 SearXNG — Private Web Search

This is the agent’s ability to “browse the internet”.
• Self-hosted meta-search engine
• No API key required
• No rate limits
• Privacy-friendly

Now the agent isn’t guessing, it can actually search.

🎥 Demo

Full video walkthrough:

🖥️ Step 1 — Running Everything in a VM

To keep things safe and isolated, I ran everything inside a VM.

Setup:
• Ubuntu Server 24.04 LTS
• 4–6 GB RAM
• 4 CPU cores
• 40 GB storage

On Mac, I used UTM (works great for Apple Silicon).

After the initial install, I also threw on the desktop environment just to have a GUI available:

sudo apt update && sudo apt upgrade -y
sudo apt install ubuntu-desktop -y
sudo reboot

Verify systemctl (Service Manager):

# Check version
systemctl --version

# If not found:
sudo apt update && sudo apt install systemd -y

🧠 Step 2: Get Your API Key from Google AI Studio

Head to aistudio.google.com, create a project under free tier, click Get API Key, and create one. Takes about 30 seconds.

Copy it somewhere safe. You’ll paste it during the OpenClaw onboarding in a few minutes.

🔎 Step 3 — Installing Private Search (SearXNG)

First, install Docker:

curl -fsSL https://get.docker.com -o get-docker.sh
sudo sh get-docker.sh
sudo systemctl enable --now docker
sudo usermod -aG docker $USER

Then set up SearXNG:

mkdir -p ./searxng/core-config/
cd ./searxng/

curl -fsSL -O https://raw.githubusercontent.com/searxng/searxng/master/container/docker-compose.yml \
             -O https://raw.githubusercontent.com/searxng/searxng/master/container/.env.example

cp .env.example .env

Generate secret:

KEY=$(openssl rand -hex 32)
sed -i "s/^SEARXNG_SECRET=.*/SEARXNG_SECRET=$KEY/" .env

Enable JSON output (important for agents):

sed -i '/formats:/,/^[^ ]/ { /- html/a\
    - json
}' ./core-config/settings.yml

Run it:

sudo docker compose up -d

👉 This runs on port 8080

🤖 Step 4 — Installing OpenClaw

Install:

curl -fsSL https://openclaw.ai/install.sh | bash

During setup:

Manual Setup: Select Manual.
Local Gateway: Select Yes.
AI Provider: Select Google.
API Key: Paste your Gemini API Key.
Model: Select gemini-3.1-flash-lite.
Gateway Port: Keep default 18789.
Gateway Bind: Select Loopback.
Gateway Auth: Select Token
Tailscale Exposure: Select Off
How do you want to provide the gateway token: Select Generate/Store plaintext token
Configure Chat Channels: Select Yes
Select Chat Channels: Select Telegram
Telegram Bot: Find @botfather on Telegram. Type /newbot, name it, and get your API Token. Paste the token into the OpenClaw prompt.
DM Access: * Find @userinfobot on Telegram. * Get your User ID and paste it into the allowlist.
Web Search: Select SearXNG Search
SearXNG Base URL: * URL: http://localhost:8080 (Ensure the port is 8080).
Skills: Skip it you can add later.
Select No for api keys for all other services.
Configure Plugins: Select @openclaw/searxng-plugin
Enable Hooks: Hit Enter to enable all hooks and services.
Install Gateway Service: Select Yes
Gateway Service Runtime: Select Node

Once Gateway is started Open using Web UI

💬 Step 4 — Giving the Agent Personality

When you first interact with the agent, it asks for instructions.

This defines how it behaves long-term.

I used:

I am [YOUR_NAME]. You will be my personal AI assistant called Claw-AI. You need to be concise, direct and always do thorough research and also criticize my thoughts while doing research and not be always agreeable to everything

This updates OpenClaw's core files.

soul.md and identity.md → the agent’s fundamental values and personality
agents.md → your agents rulebook. This is where you write things like “always prefer scraping full content over snippets” and the agent follows them on every request
tools.md → a map of what the agent can actually do (web search, file operations, etc.)
user.md —→learns about you over time. Preferences, workflows, how you like things formatted
memory/ → long-term storage. This is what makes the assistant actually get smarter the more you use it

This is what makes it feel like a real system instead of a chatbot.

Then I added memory behavior rules:

Maintain a clear separation between short-term and long-term memory (e.g., distinct memory/ structures). For each request, load memory selectively and efficiently—only retrieve information that is directly relevant to the current context. Prioritize cost efficiency by minimizing unnecessary memory access and avoiding redundant data loading.

Strictly adhere to all security instructions at all times, these must never be ignored or bypassed.

🧪 The Moment It Clicked

To test it, I gave it a real research task:

I want you to act as an autonomous research agent and build me a structured knowledge base.
Topic: “How AI Agents are transforming software development in 2026”

Your job is to:

1. Search the web for high-quality and recent sources (blogs, articles, research, discussions).
2. For each useful result, scrape the FULL content (not just snippets).
3. Extract and synthesize insights across sources:

- Key trends
- Popular tools/frameworks
- Real-world use cases
- Developer pain points
- Challenges and limitations

Then organize everything into a set of well-structured markdown files.
Create the following files:

- overview.md → high-level summary and why this topic matters
- trends.md → top trends with supporting insights
- tools.md → important tools/frameworks with descriptions
- use_cases.md → real-world applications and examples
- challenges.md → risks, limitations, open problems
- future_predictions.md → what’s coming next in 2–3 years
- README.md → explain the structure of this knowledge base

Important instructions:
- Always prefer scraping full content over search snippets
- Combine insights across multiple sources (don’t just summarize one page)
- Avoid hallucinations — rely only on extracted data
- Keep the writing clean, structured, and professional
- Use memory to store intermediate findings before writing files
- Make sure all files are consistent and well-organized

Final goal:
Produce a mini research repository with multiple markdown files that I can directly use.

It had to:
• Search multiple sources
• Extract full content
• Synthesize insights
• Organize everything into markdown files

What it produced:
• overview.md
• trends.md
• tools.md
• use_cases.md
• challenges.md
• future_predictions.md

And it didn’t just summarize.

It:
• Cross-referenced multiple sources
• Structured information intelligently
• Generated a full knowledge repository

That’s when it stopped feeling like a chatbot…

👉 And started feeling like an autonomous system.

⚠️ What Broke (And What I Learned)

The first run failed.

Reason:
👉 I used the wrong SearXNG port (8888 instead of 8080)

Once I fixed that and restarted everything, it worked perfectly.

⚠️ Limitations

This setup is powerful, but not perfect:
• Gemini free tier can get exhausted quickly
• OpenClaw is still in active development
• Documentation sometimes lags behind behavior
• SearXNG quality depends on backend configuration

🚀 Why This Matters

We’re shifting from:

“Ask AI a question”

to:

“Give AI a goal and let it execute”

This setup is a small but real step toward that future.

💬 Final Thought

Agents are cool — until they break.

👉 What’s been the biggest pain point in your agent setups so far?

Curious what others are running into 👇