The problem
Hosted bot platforms (Intercom Fin, Zendesk Answer Bot) charge €0.99-€2.50 per resolution. Self-hosted handles the same workload at ~10% of the cost while keeping customer data off third-party AI infrastructure.
Recommended setup
| Agent | Hermes Agent + Telegram/Discord/web channel via custom MCP tools |
| Hardware | Hetzner CCX13 (€38/month, dedicated CPU, 16 GB RAM) — shared CX22 isn't enough for sustained traffic |
| LLM | Claude 4.5 Sonnet for quality + OpenRouter fallback when rate-limited |
How it works
Hermes Agent indexes your help docs into a RAG store on startup. Customer-facing chat widget (or Telegram bot) sends queries to the agent. Agent retrieves relevant doc chunks, generates answer with citations, escalates to human via webhook if confidence below threshold or specific keywords detected (refund, complaint, legal).
Reality check
On a SaaS with ~600 support tickets/month, this resolves about 65% without human intervention, escalates 35%. Resolution quality survey average: 4.2/5. Customer rating distribution looks similar to the human-only baseline before deployment, with much faster median response time (12 seconds vs 6 hours).
What breaks
- Account-specific questions (the agent shouldn't have access to private customer data — escalate)
- Complaints (always escalate — humans handle these better and customers expect it)
- Documentation gaps (agent will fail confidently on questions not covered)
Alternative setups
ClawRift / NitroClaw managed equivalents at €19-100/month if you don't want the ops work. Trade-off: data goes through their infrastructure.