Skip to main content
Artificial intelligence

The AI that turns incomplete tickets into tickets you can act on

Your team loses hours chasing missing information. Dianoia generates clarifying questions automatically, reads the client's sentiment, launches autonomous agents and structures your knowledge base. Less back-and-forth, more resolution.

This isn't generic AI. It's a system trained on your support operation, your real tickets and your knowledge base. It acts where it's needed before you have to ask.

Clarifying questions generated by artificial intelligence on an incomplete support ticket
Dianoia — clarifying questions

The ticket arrives incomplete. The AI asks for what's missing.

70% of support tickets arrive without the information needed to start work. The agent loses 10-15 minutes asking the client for details. With Dianoia, the AI reads the ticket and automatically generates between 3 and 8 clarifying questions tailored to the context.

The client answers them inline, right in the ticket. A progress bar shows how many questions are still open and how many are answered. Once complete, the ticket is ready for assignment with everything needed.

Example: an incomplete ticket transformed
Original ticket: "Report export isn't working for me"
Dianoia generates:
  • 1. Which export format are you trying to use (PDF, Excel, CSV)?
  • 2. Which browser and version are you working in?
  • 3. Does an error message appear? If so, which one?
  • 4. Is it one specific report or does it happen with all of them?
  • 5. When did it last work correctly?
Progress: 0/5 answered — progress bar visible to the client
  • Generates 3 to 8 questions tailored to the ticket's context
  • Progress bar showing answered versus outstanding
  • The client answers right in the ticket, no emails involved
  • Cuts the time to a productive first contact by 60%
Automatic clarifying questions generated by Dianoia, with a progress bar and the client's inline answers
POSITIVE

Happy client. Smooth interaction. No escalation needed.

NEUTRAL

Informative tone. No emotional charge. Standard follow-up.

NEGATIVE

Discomfort detected. Prioritise the reply. Head off escalation.

FRUSTRATED

Frustrated client. Churn risk. Step in immediately.

Sentiment analysis

Spot frustration before the client escalates

Every client message is analysed in real time. The AI classifies sentiment into four levels: POSITIVE, NEUTRAL, NEGATIVE and FRUSTRATED. It doesn't rely on isolated keywords but on the full conversation context and the ticket's history.

When a ticket turns FRUSTRATED, the system automatically suggests raising its priority and can notify the team lead. Frustrated clients who aren't handled in time are the ones who cancel. This AI gives you the window to act.

  • Four sentiment levels with contextual detection
  • Automatic priority suggestion based on sentiment
  • Alerts to the team lead when frustration is detected
  • History of how sentiment evolves per ticket and per client
  • Aggregated sentiment metrics by agent, team and period
Autonomous agents

AI agents that claim tickets, analyse them and act on their own

Autonomous agents are AI processes that work on tickets without human intervention. An agent can claim a ticket, analyse the problem against the knowledge base, open a pull request with the proposed fix, send reminders to the client when information is missing and update the ticket's status.

Each agent has a dashboard with its current status, the tickets it has processed, the actions it has taken and the outcomes. The human team supervises and can step in at any point, but the heavy lifting is done by the AI.

Agent: Claims ticket #4521 — analyses the client's logs — identifies a configuration error — opens a PR with the fix
Agent: Spots a ticket with no reply for 48h — sends the client a reminder — schedules a second one
Agent: Ticket with FRUSTRATED sentiment — escalates to the team lead — proposes a conciliatory reply
  • They claim and process tickets automatically
  • They open PRs, send reminders and escalate tickets
  • A status dashboard per agent with an action history
  • Human oversight always available
Dashboard of autonomous AI agents processing support tickets, with status, actions and results
Autonomous-agent metrics — completions, errors, activity and detailed logs per agent
Agent metrics

Completions, errors and logs: every agent has its history

Launching agents isn't enough. You need to know what they do, how much they resolve and where they fail. The metrics panel shows in real time the completions (tasks finished successfully), the errors (failed attempts with a cause log) and each agent's full activity log.

You can filter by period, action type and outcome. If an agent fails in a recurring pattern, the system spots it and flags it for review. Full transparency over what the AI is doing on your behalf.

Completions

Tickets processed and resolved without human intervention.

Errors

Failed attempts with root cause and a detailed log.

Activity log

Every action recorded with a timestamp and its result.

  • Real-time metrics per agent and per period
  • Automatic detection of recurring error patterns
  • Exportable logs for auditing and continuous improvement
  • Ratio of autonomous resolution versus human escalation
AI knowledge base

Upload a PDF. The AI structures it and turns it into articles.

A knowledge base shouldn't be manual work. With AI ingestion you upload a PDF or Word file and the system extracts the content automatically, structures it into sections, generates titles and tags and publishes it as articles your team and your clients can consult.

The system also analyses frequent tickets and automatically proposes new knowledge-base articles. If the same problem comes up five times in a month, the AI drafts an article with the consolidated solution and hands it to you for review.

The automatic ingestion flow
01 You upload a PDF or Word file with technical documentation or a product manual
02 The AI extracts the content, identifies sections and builds the structure
03 Articles are created with automatic titles, tags and categories
04 Optional human review before publishing
  • Automatic ingestion from PDF and Word
  • Structure, titles and tags generated by AI
  • New articles proposed from frequent tickets
  • Semantic search across the whole knowledge base
  • Autonomous agents consult the knowledge base before replying
Knowledge base with automatic AI ingestion — articles generated from PDF and Word documents

Every AI capability inside your support platform

These aren't external plugins or integrations that break. Everything runs connected, on the same data and from the same interface.

?

Clarifying questions

The AI generates 3 to 8 questions when a ticket arrives incomplete. The client answers inline and the ticket is ready.

!

Sentiment analysis

Four levels of emotional detection in real time. An automatic alert when a client turns frustrated.

>

Autonomous agents

AI processes that claim tickets, analyse problems, open PRs and send reminders without human intervention.

#

Agent metrics

Completions, errors and activity logs per agent. Full transparency over what the AI does on your behalf.

+

AI knowledge base

Upload a PDF and get structured articles. The AI proposes new ones from frequent tickets.

*

Smart prioritisation

Sentiment, urgency and the client's history are combined to suggest the right priority for each ticket.

Next step

Request an AI demo.
See how it transforms your support.

In the demo we show you Dianoia generating questions, sentiment analysis in action and autonomous agents processing real tickets. No PowerPoint decks.

No commitment. No cost. We reply within 24 working hours.

Frequently asked questions

What people ask us about AI in support

How does the AI know which questions to ask on each ticket?
Dianoia reads the ticket, works out what information is missing to start work and generates questions tailored to the type of issue. They aren't generic: they draw on the ticket's context, the product affected and the client's history. The model improves with every ticket it processes.
Does sentiment analysis work in Spanish?
Yes. The model is trained to detect sentiment in Spanish, including idioms, irony and the colloquial phrasing common in support conversations. It also works in English for teams handling tickets in both languages.
Can autonomous agents act without supervision?
Agents can process tickets and take actions autonomously, but always within the limits you configure. You decide which action types need human approval and which can run automatically. Everything is recorded in the activity log for later auditing.
Does the AI have access to my clients' data?
The AI processes ticket data to generate questions, analyse sentiment and carry out actions. The data is not shared with third parties nor used to train external models. Processing happens on dedicated instances and the data stays under your control at all times.
Does the AI knowledge base replace manual documentation?
It complements it. You can keep writing articles by hand, but the AI speeds things up by ingesting existing documents (PDF, Word) and proposing new articles from frequent tickets. The result is a knowledge base that grows on its own and stays current with less effort.
What does the AI functionality cost? Is it a separate module?
The AI is built into the support platform; it isn't a separate module with an extra fee. Every capability — clarifying questions, sentiment, agents and knowledge base — is included in the standard plan. There are no usage charges and no token limits.
Can I switch the AI off for certain ticket types?
Yes. You can set rules so the AI only acts on certain categories, priorities or clients. You might switch off automatic questions on critical-priority tickets where the information always arrives complete, or disable autonomous agents for clients who prefer human contact.
What happens if the AI makes a mistake on a ticket?
Every AI action is recorded in the ticket's log. If an agent takes a wrong action, the team can undo it and flag it as an error. The system learns from those mistakes to avoid repeating them. You can also require human approval before certain critical actions run.
AI for support: agents and sentiment analysis

Tell us your challenge. We'll propose a solution.

No commitment. Within 24 hours, you'll receive a proposal with scope, timeline and budget. No fine print.

Book a free call →