I will build ai automation agent with n8n and local llm


About this gig
Welcome to the next level of automation.
This isn't a simple chatbotits an autonomous AI business agent that works 24/7. Standard automations follow rigid rules, but my AI agents actually reason. They monitor triggers, read unstructured context, decide actions, and handle edge cases.
Powered by local LLMs (Ollama) and n8n, your agent gets persistent memory, searches company documents (RAG), and seamlessly interacts with tools like your CRM, Slack, and Email.
Why choose me? As a DevOps Engineer, I prioritize data privacy and zero recurring LLM API costs. Everything runs locally on your infrastructure. I deploy robust, containerized architectures using my proven ZeroClaw and Moltbot setups to ensure your ecosystem is secure and scalable.
My Delivery Process:
- Architecture: Map exact triggers and AI decisions.
- Infrastructure: Deploy self-hosted n8n, Ollama & ChromaDB on your VPS.
- Core Build: Configure n8n logic and custom LLM prompts.
- Memory & RAG: Connect vector databases so the AI "reads" your docs.
- Handoff: Rigorous testing, error alerts, and a full video walkthrough.
Message me today to discuss your workflow!
Get to know Anas Rhimi
Software Engineer: Full Stack, DevOps and Linux
- FromMorocco
- Member sinceJan 2022
- Avg. response time1 hour
Languages
English, Arabic, French
My Portfolio
FAQ
How is this different from basic n8n automation?
Basic n8n uses rigid "if/then" rules and breaks on unexpected data. My AI agent uses an LLM to read unstructured data (like messy emails), reason through problems, and dynamically decide the best action, handling edge cases gracefully.
What server specs do I need for this?
To run local LLMs and vector databases securely on your own infrastructure, you need a capable VPS or dedicated server. I recommend a machine with at least 16GB-32GB of RAM (or a GPU) depending on the specific AI model size we deploy.
What if the agent makes a wrong decision?
We build "Human-in-the-Loop" (HITL) checkpoints for sensitive tasks. The agent can draft a response or prep a database entry, but it will ping you via Slack or Telegram with an "Approve/Reject" button before executing the final step.
Which local AI models do you use?
I use Ollama to run high-performing, open-source models like Llama, DeepSeek, or Qwen locally. The exact model we select depends on your server's hardware capacity and the complexity of reasoning your workflow requires.
What happens if the agent breaks or crashes?
My containerized DevOps setups are built for stability. The n8n workflows include strict error-handling. If an API fails or a process crashes, the system catches the error and sends an instant alert to your Slack or Telegram.
