Zammad 7.1: Smarter Information Extraction with AI & Regex

Released on June 17, 2026, Zammad 7.1 is a focused, practical update with a single guiding theme: turning unstructured, messy incoming requests into clean, structured, actionable ticket data — without ever giving up control over where that data lives.

If Zammad 7.0 was about bringing AI agents into the helpdesk, 7.1 is about putting them to work on the most tedious job in support: data entry. Here is what actually changed, why it matters, and what you need to watch out for before you hit the update button.

The theme of 7.1: structure from the first second

Every support team knows the pain. A customer writes a rambling email. Somewhere in that wall of text is an invoice number, a product name, a serial number, and a problem that belongs in the billing queue — but none of it is structured. An agent has to read it, interpret it, and manually copy values into ticket fields before any automation can even begin.

Zammad 7.1 attacks this problem from two angles: deterministic regex extraction for predictable patterns, and AI-powered extraction for everything that doesn't follow the rules.

Information extraction with regex

Admins can now use regular expressions inside Postmaster filters to pull specific text out of incoming emails and reuse it in filter actions. That means you can automatically populate ticket attributes based on the content of an email — not just its sender or subject line.

Typical use cases:

• Extract an order or invoice number from a structured notification email and write it into a custom field.
• Capture a system or asset ID from automated monitoring alerts.
• Pull a reference code out of a templated form submission and route accordingly.

Because regex is fully deterministic, this is ideal for clearly defined, consistent patterns — the kind that machine-generated emails and form submissions produce all day long.

New AI Agent: the Text Extractor

Regex is perfect when the format never changes. Real customers are not so cooperative. The new AI Text Extractor is built for natural language — it analyzes incoming tickets and identifies information such as contact names, product mentions, or invoice numbers even when those details don't follow any consistent format.

The extracted values are written directly into ticket attributes, which means:

• Less manual copy-paste for agents.
• Higher, more consistent data quality.
• Structured, usable information from the moment a ticket arrives — ready for downstream automation, SLAs, and reporting.

New AI Agent: the Ticket Tagger

The second new agent, the AI Ticket Tagger, automatically enriches incoming tickets with relevant tags. When there's a sudden spike in "app crashes" or a steady rise in "billing" requests, automatic tagging keeps your tickets structured, searchable, and ready for trend analysis — even as volume grows.

This is where the real long-term payoff lives. Consistent tagging from day one means:

• Better search across historical tickets.
• Reliable trend detection — you spot the wave of crash reports before it becomes a flood.
• Cleaner analytics without retroactive cleanup projects.

Privacy stays at the core

A point worth emphasizing for European and regulated organizations: Zammad's design keeps control firmly in your hands. AI features can be wired to the model and deployment that matches your compliance posture — whether that's a commercial API or a self-hosted, open-weights model running inside your own perimeter.

If you want the full walkthrough of running Zammad's AI on your own servers — the Docker setup, hosting costs, and picking a local LLM so that not a single ticket touches an external API — the OpenTicketAI team covers exactly that in Zammad 7.0 On-Premise: Full AI Power Without Your Data Leaving the Server. This article focuses on a different angle: the new 7.1 extraction features and what they mean for your day-to-day data quality, rather than the underlying on-premise infrastructure.

Before you upgrade: the fine print

Zammad 7.1 is not a "click and forget" update for self-hosted instances. Read this section carefully.

• New package repositories. Self-hosted users must migrate to the new package repositories before updating. Skipping this step will break your upgrade path.
• Elasticsearch 7 is on the way out. Support for Elasticsearch 7 ends with Zammad 7.2. If you're still on ES 7, plan your migration to a supported version now — 7.1 is your warning shot.
• Deprecated es-ca locale. The es-ca (Catalan) locale has been inactivated.
• Content-Security-Policy. Review the default CSP changes if you run custom integrations or embedded widgets.

Should you upgrade?

For most teams: yes, but plan it.

• If you process a high volume of structured, machine-generated emails, the regex Postmaster filters alone can justify the upgrade.
• If your inbound queue is dominated by free-text customer messages, the AI Text Extractor and Ticket Tagger will meaningfully cut manual triage time.
• If you're still on Elasticsearch 7, treat 7.1 as the prompt to schedule your ES migration before 7.2 forces the issue.

How Softoft can help

Adopting Zammad's new AI agents is straightforward in theory and full of edge cases in practice — model selection, data-flow design, regex tuning, and self-hosted upgrade orchestration all reward experience.