I’ve been using AI assistants for years, but there was always something off about them. They forgot our conversations. They couldn’t remember my preferences. They had no idea about my ongoing projects. Every conversation felt like starting from scratch with a stranger.
I wanted an AI assistant that actually knew me.
So I built one.
Introducing AgentAdam
AgentAdam is my personal AI assistant that runs completely under my control—no cloud subscriptions, no data mining, no corporate oversight. It lives on my own hardware, connects to my WhatsApp, and remembers everything I tell it to remember.
⚡ What Makes It Special
- WhatsApp Integration – Text my AI from anywhere, just like texting a friend
- Persistent Memory – Every conversation is remembered and searchable
- Multi-Group Support – Different contexts for family, work, and personal chats
- Agent Swarms – Spawn multiple AI agents to tackle complex tasks in parallel
- Scheduled Tasks – “Remind me every morning at 7am” actually works
- Choose Your Brain – Claude, GPT-4, Llama 3.1, or any LLM you want
Why I Built This
As a developer, I’ve always been frustrated by the limitations of consumer AI tools:
- Memory Loss – ChatGPT forgets. Claude has limited context. They’re like having amnesia every conversation.
- No Integration – I wanted my AI assistant everywhere—WhatsApp, Telegram, SMS—not trapped in a web browser.
- Privacy Concerns – My personal data, project ideas, and family conversations shouldn’t be training someone else’s AI.
- Vendor Lock-In – What happens when the API changes? Or the company shuts down? Or they raise prices 10x?
AgentAdam solves all of this.
Real-World Use Cases
Here’s how I actually use AgentAdam every day:
📊 Project Management
@Adam remind me to push the AlienAntFarm update every Friday at 3pm
📝 Content Creation
@Adam write a blog post about my Super Fontan fitness journey
@Adam create a social media thread from my latest blog post
🔍 Research Assistant
@Adam research South Carolina AI grants for startups
@Adam summarize the latest news about robotics
🤖 Agent Swarms for Complex Tasks
@Adam launch a swarm to:
1. Research competitor pricing
2. Analyze market trends
3. Draft a strategy report
(3 AI agents work in parallel, share findings, and deliver a unified report)
The Technology Stack
🛠️ Built With
- Docker Containers – Each agent runs in isolation for security
- TypeScript – Type-safe, maintainable, modern JavaScript
- WhatsApp API – Direct messaging integration via Baileys
- Multi-LLM Support – Claude, GPT-4, Ollama (local), OpenRouter, Gemini
- Vector Databases – Qdrant/ChromaDB for knowledge deduplication
- Cron Scheduling – Reliable recurring tasks
Open Source & Yours to Control
AgentAdam is open source (MIT License). You can:
- âś… Run it on your own hardware
- âś… Customize it for your needs
- âś… Keep your data 100% private
- âś… Never worry about API price hikes
- âś… Add your own features and integrations
đź”— GitHub: github.com/thinknewdev/AgentAdam
Try It Yourself
Setting up AgentAdam takes about 10 minutes:
git clone git@github.com:thinknewdev/AgentAdam.git
cd AgentAdam
npm install
cp .env.example .env
# Add your API key (or use Ollama for free)
./container/build.sh
npm run dev
Scan the WhatsApp QR code, and you’re live. 🚀
Final Thoughts
AgentAdam isn’t just a tool—it’s a philosophy:
Your AI assistant should work for you, not for a corporation.
It should remember your projects. It should respect your privacy. It should run on hardware you control. And it should never forget that conversation you had last Tuesday about starting a robotics company.
That’s what AgentAdam delivers.
If you’re tired of AI assistants with amnesia, give it a try. Your future self will thank you.