This week's newsletter was inspired by a tweet I saw from Zapier's founder, Wade
Otter auto-solved 1,000+ support tickets.
Using Zapier AI and one employee.
Allen Lai, Otter’s Head of CX:
- Routed & prioritized 10k+ tickets using sentiment analysis
- Cut down backlog by auto-resolving common requests
- Automatically flagged critical cases (like press— #Wade Foster (#@wadefoster)
4:10 PM • Jun 18, 2025
There's a case study here if you want to go deeper.
AI is fantastic at categorising messages - making it totally possible to build a customer support machine with Zapier.
I'll run through a quick guide below - just keep in mind that this works whether you're using a ticket system like Zendesk, a simple form on your website, or even just getting email inquiries - just switch out the triggers accordingly.
For my example, I've mocked up a basic Tally form and did a test submission.
Next, add an "AI by Zapier" step. A prompt builder will pop up where we can do the following
Set input fields that will be fed to the AI - in this case, the customer's message. If your form is more comple,x feed more fields in.
Set a prompt - Here, we want to analyse sentiment and summarise the help needed. Expand as needed.
Set output fields - Here, we want the sentiment and we give 3 set options, then the summary. Again, expand as needed.
Zapier will then give you a preview of the output. Looks good!
From here, you can take the zap in a few directions.
With AI analysing sentiment, you could use the Zapier Paths feature to take different actions depending on whether a customer is happy, neutral or unhappy.
For example, for happy and neutral customers, maybe a Slack message is enough. But for unhappy customers, I want to receive an SMS so I can jump on the problem quicker.
I'll leave it here, but if you think this solution would work for your business and you want help customising, just reply to this email and I'll get back to you.