Independent AI consulting & implementation, now taking on new projectsBook your free call
Solutions · IT & service desk

AI for IT and the service desk

Answer routine questions from your own runbooks and hand people well-described tickets, while every change stays with IT.

Fig. 01Example: Internal service desk assistant
  1. 01Trigger Question in chat A colleague asks for help.
  2. 02Lookup Search runbooks Your wiki and knowledge base.
  3. 03AI step Answer step by step With links to the source.
  4. 04Rule Resolved? If not, or if it is sensitive, hand off.
  5. 05Action Open a ticket With the conversation attached.

Fig. 01 · Example workflow · illustration, not client work

§01What changes

Routine questions answered on the spot, from your own runbooks.

What changes on MondayIT stops answering the same question twice and starts fixing real problems.

Today

  • Answer the same access and setup questions daily
  • Triage tickets that are missing half the details
  • Check every access request against the rules
  • Write incident timelines from memory

With the agent

  • Instant answers in chat, linked to the runbook
  • Tickets arrive complete, conversation attached
  • Access requests pre-checked and routed to the approver
  • Incident timelines drafted from the channel
§02What can be built

Agents and workflows for IT and the service desk

Each one is shaped to your own systems, rules and volumes, and goes live with your team in control.

Smaller build

Internal service desk assistant

Answers routine IT questions in chat from your runbooks, and opens a ticket with the conversation attached when it cannot solve the problem.

ChatKnowledge baseTicketing

A person decides: Any access grant or system change is done by IT staff.

Smaller build

Access request handling

Collects the details of an access request, checks it against your role rules, and sends it to the right approver.

ChatTicketingIdentity provider

A person decides: Every grant is approved by a person and applied by IT.

Smaller build

Incident timeline and summary

Builds the timeline of an incident from chat, tickets and alerts, and drafts the summary for the post-incident review.

ChatTicketingMonitoring alerts

A person decides: The incident lead owns and edits the review.

Examples of what can be built, not delivered client work.

§03Systems

Works with the systems you already use

Typical systems for it & service desk work: Chat, Ticketing, Wiki / knowledge base, Identity provider, Monitoring alerts. If yours has an API, a database or reliable exports, it can usually be connected.

  • Google Chat
  • Zoom
  • Jira
  • Asana
  • Linear
  • Trello
  • ClickUp
  • Zendesk
  • Intercom
  • Help Scout
  • Google Drive
  • Dropbox
  • Box
  • Notion

Examples of tools Georgy Consulting connects to. Logos are trademarks of their owners; no partnership or endorsement is implied. See the integrations directory.

§04Control

Where a person decides

Approval points are agreed in writing before anything is built. The signal diamond marks them in every diagram on this site.

A person approves

  • Every access grant
  • Every system change
  • Incident reviews before they are shared

Built-in controls

  • No admin permissions for agents
  • Answers only from your documents, with links
  • Existing permissions respected
§05The engagement

What an engagement delivers

Every engagement is scoped and quoted in writing after the first call. You decide at each stage whether to continue.

  1. 01

    Assessment

    A ranked shortlist of the workflows worth automating, with risks and effort stated plainly.

  2. 02

    Design

    A written design: data used, the model and provider, where people approve, and how success is measured.

  3. 03

    Build and test

    A working system in your accounts, tested on your own examples, with the results shared.

  4. 04

    Handover

    Documentation, a runbook, an access list and training, so your team can run and change it.

§06FAQ

Questions teams ask

More on the FAQ page.

Will it change settings or grant access?

No. It collects details and prepares tickets; IT staff make every change.

How does it know our setup?

From your runbooks and wiki. If the documentation is thin, improving it is part of the work.

What happens when it can’t help?

It opens a ticket with the conversation attached, so nobody has to ask the same questions twice.

§07Next step

Find your first AI win in your service desk

Book a free 20-minute call. You'll leave knowing exactly where AI pays off in your work, and what to do first.