Browse categories
Explore
Fiverr Pro
English
$
USD
I will setup n8n openclaw hermes agent hetzner ollama deepseek autogen mcp whatsapp bot
Kuba Kuch
About this gig
setup n8n openclaw hermes agent hetzner ollama deepseek autogen mcp whatsapp bot
Your whatsapp bot has no real brain behind it and your agent stack costs more than it should
Deploy your hermes agent or openclaw on a hardened hetzner vps you fully control
Connect your whatsapp business number through meta cloud api and n8n webhooks
Run your ollama local models qwen llama or mistral for cheap or free inference
Configure your deepseek v3 as a high quality low cost reasoning brain
Build your autogen multi agent conversation flow for complex task handling
Add your custom mcp servers so your agent reaches internal tools and databases
Set your tiered model routing ollama for simple tasks deepseek for real reasoning
Fix your token waste heartbeat cost bloated sessions and runaway loop bugs
Add persistent memory using memory md mem0 letta or zep across sessions
Deliver your full self hosted stack with docker and a complete walkthrough
Built for buyers who want a whatsapp bot backed by a real self hosted multi agent brain not a scripted reply system
Message me before ordering so I can scope your exact stack and budget
Get to know Kuba Kuch
Kuba Kuch
AI Automation for Businesses
- FromPoland
- Member sinceJun 2026
- Avg. response time3 hours
Languages
Polish, English, German, Russian
AI Receptionist and Appointment Booking solutions for local businesses that want to capture more leads, answer calls 24/7, and automate scheduling. I build AI voice agents using Vapi, Retell AI, Bland AI, OpenAI, ElevenLabs, Twilio, GoHighLevel, n8n, Make, and CRM integrations to streamline customer interactions. Message me with your business workflow or automation goal, and I'll tell you exactly what I can build for you.
My Portfolio
FAQ
Is this a real multi agent system or just one bot answering whatsapp?
Yes it is a real multi agent system. I configure autogen conversation flows alongside hermes agent or openclaw so different agents handle research, decisions, and replies, instead of one script matching keywords to canned answers.
Will this actually cost less than a claude or gpt only setup?
Yes. I route simple whatsapp replies through ollama running locally for free, and use deepseek v3 for real reasoning at a fraction of claude or gpt pricing, so your monthly bill drops without losing response quality.
Can this connect to my whatsapp business number directly?
Yes. I connect your whatsapp business number through meta cloud api and n8n webhooks, so incoming messages trigger your agent stack and replies send back automatically.
Do I need my own hetzner account or do you provide the server?
Yes you need your own hetzner, contabo, or similar vps account since you own and control the infrastructure directly, and I handle the full setup, hardening, and deployment on it.
Will you fix token waste and loop bugs if my current setup already has them?
Yes. I audit heartbeat frequency, session bloat, and runaway loop causes, then configure tiered model routing and spend caps so the same expensive mistakes do not repeat.
Can this reach my internal tools or databases beyond whatsapp?
Yes. I build custom mcp servers so your agent stack can read and write to internal tools, apis, or databases that have no existing integration, not just handle whatsapp replies.
Do I get a walkthrough so I can manage this myself after delivery?
Yes, every delivery includes a full walkthrough covering how the model routing and multi agent flow works, and how to monitor cost and performance yourself.
