Support

Extract Support Ticket Metadata with AI

Product, issue summary, stated severity, steps already tried — ticket fields extracted from free-text customer messages, without the model's own judgment leaking in.

Open in Extraction Prompt Generator

Overview

Customer messages arrive as stories; support queues need fields. This setup extracts customer_name, product, issue_summary, severity, steps_taken, and reply_needed — with the extraction-not-judgment line drawn explicitly: severity copies the urgency the customer stated ("blocked", "urgent"), it is never the model's own assessment. The issue summary stays faithful to the customer's wording in at most two sentences, steps_taken becomes a deduplicated list, and reply_needed answers true or false from the text alone. Strict policy plus null discipline keeps the queue honest.

How to use this resource

Generate and pipe messages through

Each customer message becomes one record with the same six fields — the queue stays uniform.
Respect the severity boundary

The rule says "use the level stated in the text — do not substitute your own judgment." Re-grading happens later, by your rules.
Use steps_taken in the first reply

A deduplicated list of what they tried means the first response never opens with advice they've already followed.

Why This Works

Extraction-not-judgment framing keeps the model from playing support engineer
The faithful issue summary preserves the customer's words for the human who replies
reply_needed as text-grounded true/false makes auto-acknowledgment flows safe

Best for

Support teams structuring free-text channels into a ticketing system
Queues where "what did they try" saves the first reply cycle
Pipelines that must distinguish stated severity from assessed severity

Not for

Assigning the ticket's category or final priority — that's the Data Classification Prompt
Writing the response — extraction prepares the ticket, not the answer

Use cases

Converting inbox messages into structured queue entries
Capturing what the customer already tried as a clean list
Recording stated urgency without the model re-grading it

FAQ

Can the AI assign its own severity to a support ticket here?

It cannot — the severity rule says 'Use the level stated in the text — do not substitute your own judgment,' so severity copies the urgency the customer literally wrote (like 'blocked' or 'urgent') and never the model's own grade. If the customer states no urgency, MISSING DATA sets severity to null. Re-grading against your own priority scale is a separate later step.

What happens to fields the customer message never mentions, like customer_name or steps_taken?

Any field not found in the source is set to null and its key is still returned — 'do not omit the key' — so every extracted ticket carries the same six keys and your queue rows stay uniform. Values that are only implied but not stated are also treated as missing under the AMBIGUITY POLICY, never guessed or invented.

How do I change which ticket fields get extracted from customer messages?

Edit the FIELDS TO EXTRACT and EXTRACTION RULES sections through the extraction-prompt-generator, which outputs a revised prompt you run in your own assistant. You can drop steps_taken or add a field, but keep the reply_needed rule ('answer true or false based only on what the text states') text-grounded if auto-acknowledgment flows depend on it. NewPrompt generates the prompt, not the extraction itself.

Customize This Resource

Opens this setup in Extraction Prompt Generator. Generate to get the full extraction prompt — then adjust the fields, missing-data behavior, and ambiguity policy.

Open in Extraction Prompt Generator

Prompt Template

Copy it as-is, or use Open in Extraction Prompt Generator to load it pre-filled and customize it with your own context.

TASK
Extract ticket metadata from the customer message in the input for our support queue.

SOURCE
The input is a customer support message describing a problem.
Reading guidance: Urgency words in the customer's own text ("urgent", "blocked") matter — extract what is stated, not your own judgment.

FIELDS TO EXTRACT
- customer_name (optional): Customer's name if given
- product (required): Product or feature the message is about
- issue_summary (required): The problem in one or two sentences, faithful to the customer's wording
- severity (optional): Urgency level stated by the customer
- steps_taken (optional): What the customer already tried
- reply_needed (required): Whether the customer expects a reply

EXTRACTION RULES
- customer_name: The person's full name as written — do not expand initials or invent missing parts.
- product: The product's display name as written.
- issue_summary: Faithful to the source wording, no added interpretation — at most two sentences.
- severity: Use the level stated in the text — do not substitute your own judgment.
- steps_taken: Extract every distinct item as a separate list entry; deduplicate near-identical ones.
- reply_needed: Answer true or false based only on what the text states.
- Extract values from the source text only — do not use outside knowledge to fill fields.
- Do not summarize or interpret the document — return only the fields listed above.

MISSING DATA
- If a field is not found in the source, set it to null — do not omit the key.
- Never invent or guess a value for a missing field.

AMBIGUITY POLICY
- Extract only information explicitly stated in the source text — do not infer values.
- If a value is implied but not stated, treat the field as missing.

OUTPUT FORMAT
Return the extracted data as a single valid JSON object using the field names below.

EXAMPLE OF A VALID EXTRACTION
{
  "customer_name": "Jane Smith",
  "product": "Aurora Desk Lamp",
  "issue_summary": "Short description taken from the source",
  "severity": "high",
  "steps_taken": [
    "Restarted the app",
    "Cleared the cache"
  ],
  "reply_needed": true
}

OUTPUT RULES
- Return only the JSON object — no text before or after it.
- Do not wrap the output in markdown code fences.
- Use the field names exactly as defined — no renaming, no extra fields.

More resources from Extraction Prompt Generator

Resource

Extract Data From Text with AI

Free text in, named fields out. The extraction prompt pattern that turns any unstructured text into consistent, parseable records.

Prompt Engineering

Resource

Extract Invoice Data with AI

Invoice number, vendor, dates, total, currency — extracted into clean fields with strict no-inference rules, ready for accounts payable.

Operations

Resource

Missing Data in AI Extraction — Null, Unknown, or Skip

The most consequential setting in any extraction prompt: what the model does when the field isn't in the text. Four behaviors, and when each is right.

Prompt Engineering

Resources that pair well

Resource

Force JSON Output from AI

Stop getting 'Sure, here is the JSON…' — the output-contract pattern that forces models to return only parseable JSON: schema, example, and a strict rule block.

Prompt Engineering

Resource

Classify Support Tickets with AI

Billing, Technical, Account, How-To, Feature Request — ticket triage with definitions that decide the borderline cases for the model.

Support

Resource

Fix Invalid JSON from AI

The JSON won't parse and you can't see why. Deterministic cause-sniffing — trailing commas, single quotes, unclosed brackets — and the repair prompt that fixes it.

Engineering

Related tools

Tool

Extraction Prompt Generator

Build prompts that extract defined fields from unstructured text — emails, invoices, tickets, résumés.

Structured Output

Workflows that use this resource

Workflow

AI Customer Support Workflow

Run inbound support the same way every time — triage and route the ticket, pull the details that matter, draft a reply in a consistent voice, and log the resolution for the record.

4 steps 20–40 minutes

Tip: Save time by exploring related resources and tools that integrate with this resource.