Self-hosted customer support AI chatbot (Dify)
Quick answer
Dify lets you build a chatbot that answers from your help-centre documents and publish it as a web app that you can embed on your website. LiteLLM is optional: it gives you one gateway, budgets and spend logs across model providers. This covers the AI answer bot only, not a helpdesk with tickets and inboxes. Running Dify on one Hetzner Cloud server starts at about €19.99/month (CPX22, excl. VAT). Model API usage, backups and your time come on top.
Last updated · sizing from each app's official docs · prices from Hetzner Cloud (2026-10-22)
Components
- Dify: Chatbot and knowledge-base builder, web app publishing Official hardware: minimum 2 vCPU · 4 GB RAM · License: Dify Open Source License (modified Apache-2.0) (Custom / conditional)
- LiteLLMoptional: Model gateway with budgets and spend logs Official hardware: minimum 1 vCPU · 4 GB RAM · License: MIT (core) + LiteLLM Enterprise License (enterprise directory) (Open core)
Combined hardware and monthly cost
| Required components, official minimums | 2 vCPU · 4 GB RAM (sum of official minimums of Dify) |
|---|---|
| Smallest server that fits | CPX22 (2 vCPU, 4 GB RAM, 80 GB disk) €19.99/mo |
| Including optional components | 3 vCPU · 8 GB RAM (sum of official minimums of Dify + LiteLLM) → CPX32 €35.99/mo |
How we got this: we added up the official minimum vCPU and RAM each project publishes (linked on its app page) and took the cheapest Hetzner Cloud Regular Performance (shared AMD vCPU) plan that meets the total. Prices read from www.hetzner.com/cloud/regular-performance on 2026-10-22. "From" monthly price cap shown by Hetzner, excluding VAT. Prices vary by location. Covers the server only; it excludes LLM API usage, backups, extra storage, GPUs and your time. Running everything on one server is our simplifying assumption. Measured figures, where we have them, are on each app page.
Cost vs Intercom Fin / Zendesk AI
Assumptions and sources
Deployment outline
- Clone the Dify repo, copy .env.example to .env and start it with Docker Compose (v2.24.0+), then open /install. source
- Configure a model provider in the workspace settings. source
- Build a chatflow on your knowledge base, publish it and embed it on your website. source
This is a summary of the official docs, not a full tutorial. Each app page has the exact official install command.
Gotchas
- Dify: Needs Docker Compose v2.24.0 or later. source
- Dify: The license forbids operating a multi-tenant (multi-workspace) service or removing the Dify branding from the web console without a commercial license. source
- LiteLLM: Without a database, budgets are not enforced. max_budget is silently ignored apart from a one-time startup warning, and virtual keys do not work. source
- LiteLLM: Memory is not a usable autoscaling signal for the proxy; scale on CPU instead. source
What this does not replace
- None of these is a helpdesk. There is no shared inbox, ticketing, SLA or human-handoff workflow unless you integrate one.
- You connect your own model (API or self-hosted), so per-conversation model costs still apply.
- None of these replaces Zendesk's ticketing, omnichannel inbox, voice or reporting. They cover the AI answer bot only.
- You connect your own model (API or self-hosted).
FAQ
Can I offer a Dify-based chatbot to several clients?
Dify's license forbids running a multi-tenant (multi-workspace) service without written permission. One company running its own bot is fine; an agency hosting many clients should read the license or buy a commercial one.
How much does a customer support ai chatbot stack cost per month?
Running Dify on one Hetzner Cloud server starts at about €19.99/month (CPX22, excl. VAT). Model API usage, backups and your time come on top.
What hardware does Dify need?
2 vCPU · 4 GB RAM (sum of official minimums of Dify).
Can a business use these apps commercially?
Dify: Dify Open Source License (modified Apache-2.0) (internal business use allowed). LiteLLM: MIT (core) + LiteLLM Enterprise License (enterprise directory) (internal business use allowed). Summaries only, not legal advice.
Is it cheaper than Intercom Fin or Zendesk AI?
It depends on team size, API usage and how you value your time. Use the calculator on this page; it starts with 10 users and Intercom Fin Essential (includes Fin AI Agent) at the price published on 2026-10-22.
Other stacks
- Private ChatGPT for a small business (Open WebUI + LiteLLM)
- Private document Q&A over your company's knowledge (Onyx)
- Self-hosted AI workflow automation (a Zapier replacement with n8n)
- Self-hosted meeting transcription (Speaches + Open WebUI)
- Self-hosted coding assistant for your dev team (Tabby)
- Fully offline private AI chat (Ollama + Open WebUI)
Every figure comes from the linked app pages and official docs. Spotted something wrong? Tell us.