# Build a customer-support agent

> Which AI agent should I use for customer support?

Support agents resolve tickets across chat, email, and voice. This category is commercial-heavy — resolution quality, integrations, and escalation flows are the product — but there's a credible open-source route if you want the knowledge base on your own servers.

A hand-picked shortlist from the Agent Search Engine index. Every pick is a real listing; the adoption, pricing, and activity figures are the index's live numbers. Browse on the web: https://agentsearchengine.app/use/build-customer-support-agent

## Recommended picks

### Easiest if you're on Intercom: Fin by Intercom
- Paid product · Agent
- **Setup:** Managed — add-on
- **Why:** Drops into an existing Intercom setup and starts resolving a share of conversations — the shortest path if that's already your help desk.
- [listing](https://agentsearchengine.app/agents/intercom-fin) · [official](https://fin.ai)

### Enterprise-grade: Sierra
- Paid product · Platform
- **Setup:** Managed — sales process
- **Why:** White-glove conversational AI for customer experience at enterprise scale — built around brand-safe resolution rather than a self-serve toolkit.
- [listing](https://agentsearchengine.app/agents/sierra) · [official](https://sierra.ai)

### Self-hosted option: MaxKB
- Open source · Platform · 23k★ · Python · GPL-3.0 · active 3 days ago
- **Setup:** Self-host — Docker
- **Why:** The most adopted open-source knowledge-base agent in the index — your docs, your servers, no per-resolution pricing. You trade vendor polish for control.
- [listing](https://agentsearchengine.app/agents/maxkb) · [official](https://maxkb.cn)

## What to watch for
Support agents fail publicly — a wrong answer goes straight to a customer. Whatever you deploy, start with a narrow scope of ticket types, keep confident escalation to humans, and measure deflection honestly before widening it.

---
Source: [Agent Search Engine](https://agentsearchengine.app) — more task-first guides: https://agentsearchengine.app/use