Support Analyst

Ask questions about your tickets in plain English and get tables, summaries and trends back. Covers search, sentiment, deep dives, analytics, QA and alerts.

Support Analyst is an AI analyst for your ticket data. Ask a question in plain English, such as “show me all open negative-sentiment tickets from last week”, and get a table, summary, timeline or trend with links to the source tickets.

It uses the ticket data IrisAgent already ingests from Zendesk, Salesforce, Intercom, Freshworks, HubSpot or Jira Service Desk. No extra setup is needed.

​

How it works

1. Ask a question in the Support Analyst chat.
2. Support Analyst searches, aggregates and reads your ticket data.
3. You get an answer with links back to the tickets.

Ask follow-ups in the same conversation. It keeps context, so you can drill from a trend down to one ticket:

“How many tickets did we close last week?” “Of those, how many had negative sentiment?” “Show me the top 5 by ticket age.”

Conversation memory lasts 24 hours per session.

​

What you can ask

Capability	What it does	Example
Search	Filter by keyword, status, priority, assignee, account, sentiment, category and date. Searches ticket text, comments, subject, tags and metadata.	”Open tickets from Acme Corp where the customer is frustrated”
Ticket deep dive	Full timeline of one ticket: who said what and when, current status, and how it evolved.	”Summarize ticket 4821”
Sentiment	Query by sentiment: Negative, Moderately Negative, Neutral, Moderately Positive, Positive.	”Accounts with the worst sentiment in the last 30 days”
Analytics	Volume by status, priority, assignee, account, sentiment, category, module or tag. Trends over time, top N lists, response times, agent performance.	”Ticket volume by priority this month vs last month”
Account overview	Open and total tickets, recent activity, top categories, average sentiment and a health indicator.	”Give me an overview of Acme”
Conversation QA	Grades a ticket against your QA rules and flags the comments that failed.	”QA ticket 4821”
Answer grounding QA	Grades a delivered AI answer against the articles it cited. See below.	”Check the grounding of the AI answer for case 12345”
Proactive alerts	Finds spikes, new issue patterns and recurring complaints across accounts.	”Any new issue patterns in the last 24 hours?”

​

How analytics counts

• Status questions use the ticket’s current status. There is no history of when a ticket changed status. Date ranges filter on ticket creation time.
• Module and tag breakdowns show only the top groups. Missing values are reported as an unset count. If Module is not populated, the answer says so.
• Tag counts can exceed the ticket total (a ticket can have several tags), so they are not shown as percentages.
• Agent filters match display name, email or a full-name email alias. A partial name works only if it matches exactly one agent.

​

Answer grounding QA

Ask “check the grounding of the AI answer for case 12345”, or use the Validate grounding skill. Provide exactly one of: a case number, a source case ID or a chat ID. It works for case answers and chat answers.

It checks two things:

• Retrieval relevance: were the cited articles about the question?
• Grounding faithfulness: is every claim supported by those articles?

It returns correct or defect, and for a defect, the type:

Type	Meaning
Hallucination	A claim is not in any cited article (a generation problem).
Wrong article	The answer is faithful to an article that does not address the question (a retrieval problem, not a pass).
Incomplete	Grounded and relevant, but missing a critical step from the cited article.
KB gap	No cited article contains the answer (a knowledge-base coverage problem).

This is an on-demand check. An always-on audit of every AI answer is on the roadmap.

​

Voice and other languages

Submit an audio transcript with your query and Support Analyst uses it as primary context. Non-English transcripts and ticket text are translated before analysis, so you can search and run QA across languages from one English interface.

​

Where to use it

• IrisAgent dashboard: the Support Analyst chat.
• API: POST /v1/ticket/query, for Slack bots, internal dashboards or scheduled reports. See the API reference.
• Audio API: POST /v1/ticket/query/audio for voice transcripts.

​

Data sources

The more sources you connect, the more it can answer:

• Ticketing systems: Zendesk, Salesforce, Intercom, Freshworks, HubSpot, Jira Service Desk
• Knowledge bases: Confluence, Public websites, Uploaded content
• CRM: Salesforce CRM

​

Security and privacy

• Queries are scoped to your account. Support Analyst never returns other customers’ data.
• IrisAgent is SOC 2 Type II certified and GDPR compliant.
• Data is encrypted at rest and in transit.
• Conversation history is kept for 24 hours per session and is not used for cross-customer model training.

​

FAQ

Do I need to set anything up? No. Once IrisAgent has ingested your tickets, Support Analyst is ready.

Can it reply to, update or escalate tickets? No. It is read-only today. Actions are on the roadmap.

How fresh is the data? It uses the same indexed data as the rest of IrisAgent. In most cases new tickets and comments appear within minutes, depending on your integration.

Which AI model does it use? A configurable LLM (OpenAI, Anthropic or Google). Enterprise customers can request a model preference.

Does it hallucinate? Answers are grounded in your indexed ticket data and include ticket IDs and links so you can check them. If data is missing, it says so instead of guessing.

Want a walkthrough? Email us.