Customer support · Production

Assembled dogfoods Cal AI assistant to accelerate support onboarding and agent productivity

The problem

Support agents and rotation participants lacked a fast way to find answers and draft responses; new hires faced ramp times of six or seven months before independently handling tickets; non-support staff on rotation were entirely unfamiliar with where to start on technical questions.

Workflow diagram · grounded in source
1
Ticket arrives in Zendesk
trigger
“Cal was accessible directly in Zendesk”
2
Cal searches documentation
ai_action
“a simple Slackbot that would search support documentation to quickly find answers to questions asked over Slack”
3
Suggested reply with cited sources
output
“it cites the sources it uses in its suggested replies”
4
Agent reviews and responds
human_review
“when a support person can rely on an AI-powered research assistant to check the help center, it makes them work so much faster and better. This frees them up to really dig into the issue while creating a good experience for the customer.”
5
Documentation gap feedback
feedback_loop
“We've updated plenty of support articles because, through the process of using Cal to troubleshoot an issue, we've identified pertinent missing information”
Reported outcome

A new support engineer answered all tickets by the second or third month—described as 2–3× faster onboarding compared to a previous six-or-seven-month ramp.
Rotation participants felt more informed and self-sufficient. Agents work so much faster and better overall, and documentation gaps were identified and fixed as a side effect of Cal usage.

Reported metrics
New hire time to answering all ticketssecond or third month
Previous-role ramp time (baseline)six or seven months
Onboarding speed improvement2–3× faster
Agent work speed and qualityso much faster and better
Show all 6 reported metrics
new hire time to answering all ticketssecond or third month
previous-role ramp time (baseline)six or seven months
onboarding speed improvement2–3× faster
agent work speed and qualityso much faster and better
support rotation improvementbig improvement
productivity multiplier for new usersforce multiplier early on
Reported stack
CalZendeskSlack
Source
https://www.assembled.com/blog/observations-from-dogfooding-our-own-ai-product
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

A new support engineer answered all tickets by the second or third month—described as 2–3× faster onboarding compared to a previous six-or-seven-month ramp.

What tools did this team use?

Cal, Zendesk, Slack.

What results were reported?

New hire time to answering all tickets: second or third month; Previous-role ramp time (baseline): six or seven months; Onboarding speed improvement: 2–3× faster; Agent work speed and quality: so much faster and better (source-reported, not independently verified).

How is this customer support AI workflow structured?

Ticket arrives in Zendesk → Cal searches documentation → Suggested reply with cited sources → Agent reviews and responds → Documentation gap feedback.

Related customer support cases