/ connectors · Zendesk

Connect Zendesk to Mari

Your Help Center is the knowledge your team already decided customers can rely on. It's vetted, published, and maintained. And internally, almost nobody searches it: support answers from memory, engineering doesn't know it exists, and the two drift apart quietly.

Mari's Zendesk connector puts your published support knowledge in the same searchable base as everything else your team knows.

What syncs

• Help Center articles: the published, vetted knowledge. Tickets are deliberately skipped, since the articles are the knowledge and the tickets are the queue.
• Incremental sync through Mari's one pipeline: chunk, content-hash, embed. Unchanged articles are never re-embedded.

Connecting takes a minute

Three fields: your subdomain (the yourco in yourco.zendesk.com), the agent email the token belongs to, and a Zendesk API token. Test connection shows Zendesk's actual error if something's off, verbatim. Connect, and a scheduled sync flow keeps the Help Center fresh from then on.

What you get back

Ask a question and the answer streams with numbered citations. When the answer exists in a published article, that's what gets cited, so internal answers match what you've already told customers. Help Center content ranks alongside your repos, Slack threads, and wikis in one hybrid search. Coverage gaps show the questions your team actually asked that no approved answer covers, which is a running list of what to write next.

Runs on your infrastructure

Mari is self-hosted and Apache 2.0. Postgres with pgvector is the whole data plane, models run locally through ollama, and every LLM feature has a deterministic fallback. Your support knowledge never leaves your deployment.

Start here

• Quick start: docker compose up to a working knowledge base.
• The docs cover connecting a source step by step.
• Book a demo to see it live.

Also connects: GitHub, Slack, Confluence, Jira, Linear, and more. 12 connectors, one pipeline.