Customer storyCactus Club Cafe logo

Tier 1 tech support for a restaurant group, drafted from the runbooks the team wrote

Cactus Club Cafe runs its internal IT helpdesk in Zendesk. Tixie, the tier 1 agent, reads the whole ticket, opens the emergency policy and the routing matrix, and drafts the fix from the matching runbook, step for step, with nothing improvised.

Industry
Restaurants, internal tech support
Based in
Vancouver, Canada
With eesel since
April 2026
Working with
Gerardo Moran, Manager, Tech Support at Cactus Club Cafe
750Zendesk tickets handled in a month8 Aug to 8 Sep 2026
790draft replies for the tech support teamagainst 51 sent directly
1,650runbook and policy lookupsabout two documents read per ticket
2agentsTixie on tickets, an audit agent on the daily finance extract
cactusclubcafe.com
Cactus Club Cafe website
How it runs

What happens when a customer writes to Cactus Club Cafe

  1. 1
    A restaurant raises a ticket

    Location staff write to the Zendesk tech desk about their systems.

  2. 2
    Read everything, then the two gate documents

    Tixie reads the history, emergency policy and routing matrix first.

  3. 3
    Match the runbook, follow it exactly

    It sets the type and drafts the runbook steps as written, unchanged.

  4. 4
    One combined draft, then stop

    A single reply for the team to send, with no sign off.

Connected integrations

  • Google Drive
  • Zendesk

Agents at work

  • Tixie | Tier 1 Agent807 runs in 30 days
  • Extract Agent60 runs in 30 days

The company

Cactus Club Cafe is a Canadian restaurant group headquartered in Vancouver, with the King Taps brand alongside it. Its tech support team looks after the systems every location runs on, and the restaurants reach that team through a Zendesk helpdesk.

Visit Cactus Club Cafe at cactusclubcafe.com

The problem

Tier 1 tech support is mostly known problems with known fixes, and the fixes are written down. The cost is in reading the whole ticket, finding the right runbook and typing the steps out again, at the pace restaurants raise them. Gerardo Moran, Manager of Tech Support, wanted an agent that would do exactly that and nothing more: no improvised troubleshooting, no guessing when a document is missing, and no changes to a ticket the instructions did not call for.

How it runs

Cactus Club Cafe connected eesel to Zendesk and Google Drive in April 2026. The team wrote Tixie's core instruction and a library of referenced documents, and the instruction makes every reference mandatory.

  • Read first. Tixie calls the full ticket with comments and reads it from first to last. It acts only on the latest requester update; earlier comments are context.
  • Two gates before any tool call: the emergency policy and the conditional execution matrix. If the ticket is an emergency, the policy's response goes out and Tixie stops.
  • Runbooks are followed verbatim. The instruction forbids adding, removing, summarising, reordering or improvising technical steps from a matched document.
  • Tixie sets the ticket type, and otherwise never closes a ticket or changes tags, status, priority, group or assignee unless a document says to.
  • The team can ask for a draft on any ticket by leaving an internal note that says @eesel draft a response.
  • A second agent audits the daily finance extract. Every morning a PDF arrives from the reporting system; the agent files it, checks each store pulled, compares the grand total debit and credit, and either solves the ticket as balanced or leaves it open with the imbalance and the store named.

A month of work

Between 8 August and 8 September 2026 Tixie handled about 750 Zendesk tickets. It wrote about 790 draft replies for the team and sent 51 directly, left 146 internal notes, tagged 82 tickets and routed 39. It read runbooks and policies about 1,650 times, roughly two documents per ticket. The extract audit ran about 60 times, once a morning plus a catch up run.

What is next

Gerardo has been shaping the setup week by week: filters so Tixie only wakes on the right groups and tags, a rule that it should take no action on a ticket if it cannot read its instructions, and a plan for an intake agent that routes to specialist agents by task. The knowledge base is about 4,200 Zendesk items plus the runbook files in Drive.

A month of work

What eesel did, from its own records

Every count on this page comes from eesel's activity log for Cactus Club Cafe, 8 Aug to 8 Sep 2026. The only estimate is the money and time saved, and the receipt shows how it is worked out.
$1,900
of people’s time a month, $23,000 a year
75 hours
about a working week, every month

$25 is an international benchmark for an hour of support time, not Cactus Club Cafe’s own cost.

Conversations handled, 8 Aug to 8 Sep 2026750
×Minutes a person would spend on one6
=Hours of people’s time75
×Cost of an hour, all in$25
=A month$1,900
A year$23,000
Zendesk tickets and other live work
Live work per week, the last 13 weeks
0121242Week of 15 Jun: 242Week of 22 Jun: 19022 JunWeek of 29 Jun: 152Week of 6 Jul: 1566 JulWeek of 13 Jul: 201Week of 20 Jul: 17320 JulWeek of 27 Jul: 166Week of 3 Aug: 1243 AugWeek of 10 Aug: 177Week of 17 Aug: 19217 AugWeek of 24 Aug: 174Week of 31 Aug: 17031 AugWeek of 7 Sept: 61
Show as a table
Week ofRuns
15 Jun242
22 Jun190
29 Jun152
6 Jul156
13 Jul201
20 Jul173
27 Jul166
3 Aug124
10 Aug177
17 Aug192
24 Aug174
31 Aug170
7 Sept61
When the work arrives
Hour of day, UTC. 20% of it lands between 10pm and 7am, when nobody is at a desk.
00:00 UTC: 700:0001:00 UTC: 1902:00 UTC: 503:00 UTC: 904:00 UTC: 605:00 UTC: 206:00 UTC: 206:0007:00 UTC: 008:00 UTC: 109:00 UTC: 110:00 UTC: 011:00 UTC: 112:00 UTC: 612:0013:00 UTC: 1214:00 UTC: 2415:00 UTC: 6016:00 UTC: 8917:00 UTC: 10318:00 UTC: 6518:0019:00 UTC: 8720:00 UTC: 9521:00 UTC: 7822:00 UTC: 6723:00 UTC: 36
Show as a table
Hour (UTC)Runs
00:007
01:0019
02:005
03:009
04:006
05:002
06:002
07:000
08:001
09:001
10:000
11:001
12:006
13:0012
14:0024
15:0060
16:0089
17:00103
18:0065
19:0087
20:0095
21:0078
22:0067
23:0036
What it did
Actions taken in the customer's tools, last 30 days
Knowledge searchesKnowledge searches: 1,6511,651Draft replies for the teamDraft replies for the team: 792792Internal notesInternal notes: 146146Tickets taggedTickets tagged: 8282Public replies sentPublic replies sent: 5151Tickets routedTickets routed: 3939
Show as a table
ActionCount
Knowledge searches1,651
Draft replies for the team792
Internal notes146
Tickets tagged82
Public replies sent51
Tickets routed39

Hire an eesel teammate for your support

Connect your helpdesk, let it learn from your past tickets, and start with drafts before it replies on its own.

Start free