For service desk leads and MSP teams
MockDesk puts every L1 and L2 technician in front of live tickets before you hire them. An AI client answers like a real user, a simulated console answers like a real tenant, and you get a scorecard backed by evidence.
Watch it work
This is a real scenario from the library. The client says her clients got an invoice she never sent. Follow how a strong technician works it, and what the scorecard says at the end.
0 critical misses · 1 warning
Why this exists
Every service desk lead has hired someone who answered every question right, then struggled the first time a real user called with a vague problem and a deadline.
The gap is not knowledge. It is process: asking before acting, verifying identity, reading the logs, knowing when a password reset is really a security incident, and writing notes the next person can pick up.
MockDesk tests exactly that, the same way the job does. A user who only tells you what you ask for. A tenant that only shows you what you look up. A clock running.
How it works
Pick Level 1 or Level 2, how many questions and live tickets, and which systems to cover. Each person gets a random draw.
The candidate needs no account. Progress saves as they go, and the answers never reach their browser.
A short knowledge check, then live tickets with an AI client who reveals facts only when asked, and a console that answers every command.
Six rubric scores with evidence from the transcript, flags for critical misses, and the exact questions to ask in the next interview.
What's inside
Each scenario has a persona, a temperament and facts that only come out when the technician asks the right question. Vague questions get vague answers. Asking for a password gets refused.
Exchange Online, Entra, AD, DHCP and the user's PC. Real cmdlets get realistic output. Fixes change what the client sees next.
PS> Get-DhcpServerv4OptionValue -OptionId 6
OptionId Value
6 {10.0.0.12, 10.0.0.13}
Resetting without verifying identity, or disabling MFA for a senior partner, is flagged no matter the score.
Root causes, correct answers and grading prompts live on the server. Candidates only ever see the ticket.
Every generated question and ticket is shown to you first with its answer and source. Remove anything weak.
Two ways to use it
The scorecard
Every score points to something the person actually did or didn't do in the chat and the console. Read the full transcript, see every command, and download it for your records.
Knowledge 12/15 · Practical 84, 88
Plans
Coming soon
Try it with the built-in library
Coming soon
Your own tickets, small team
Coming soon
Connect your ticketing tool and KB
Coming soon
No limits
Works with Jira Service Management, ServiceNow, Freshservice and Zendesk. Any other tool works through CSV.
I lead a team of service desk technicians. I built this because I kept seeing the same thing: people who interviewed well and then froze on their first messy ticket. I wanted a way to see how someone works before day one.
Questions
No. They open one private link. Their chat, commands and notes save as they go, so a reload doesn't lose work.
No. Root causes, correct answers and grading instructions stay on the server. The browser only receives the ticket and the questions.
Microsoft 365 and Entra ID, Exchange Online, Windows and Active Directory, networking basics, and security response. Evaluations built from your own tickets cover whatever those tickets cover.
It is stored in the site's database and sent to the Claude API to write the test. Emails and phone numbers are stripped on import, names are replaced with fictional ones in what the AI writes, and connection credentials are encrypted. Only upload data you are allowed to process.
Jira Service Management, ServiceNow, Freshservice and Zendesk, with a read-only account. You pick the support groups and date range, and it pulls resolved tickets with their replies and notes, drops thin and duplicate ones, and sorts them by issue type. Anything else works through a CSV export.
Tickets show how someone did the work, not always how it should be done. With your runbooks loaded, answers and expected steps follow the documented procedure, and issue types with no article are flagged so you know what to write next.
An AI grader scores six criteria from 0 to 5 using the full transcript, the console log and the close out notes, compared against the scenario's expected path. You can read every piece of evidence and regrade.
Take the free demo: 5 questions and 1 live ticket, then see the same scorecard a hiring manager gets. About 15 minutes.
Start the free demo