Been using Lindy's free tier for a few weeks and finally got something working I'm really excited about. We use Calendly for client bookings, but the confirmation emails were just... basic. No follow-up, no reminders, no adding details to our internal spreadsheet.
So I built a Lindy agent that triggers off new Calendly events. Now when someone books, it:
1. Sends a custom confirmation with a link to a pre-call questionnaire (via a Google Form).
2. Adds the client's name, email, and meeting time to our Airtable base automatically.
3. Sends me a Slack DM with the details so I'm always in the loop.
It feels like magic? The logic is pretty simple but having it all connected saves me so much manual copying and pasting. Still tweaking it, but the free tier has been plenty for this. Anyone else doing something similar? Curious about other ways to use agents for customer touchpoints.
Nice setup. I always start calculating the manual cost of that copying and pasting you mentioned - even 5 minutes per booking adds up fast over a month.
Have you looked at what happens when you scale? The free tier is great, but Lindy's pricing page gets murky after that. I ran the numbers for a similar workflow last quarter. The jump to a paid plan can be steep if your agent count or task volume grows, which might happen if you add more steps like sending reminders.
You might want to keep an eye on your agent's execution logs. Sometimes these automations fail silently, especially when connecting to external APIs like Airtable or Google Forms, and you only notice when a booking falls through the cracks.
Scaling cost is one issue, but my bigger concern with these orchestrators is the latency tax. Each of those sequential API calls - Calendly webhook, then Airtable, then Slack, maybe Google Forms - adds a chain of synchronous HTTP requests. If any single external service has a transient slowdown, the entire agent execution blocks, and you risk hitting timeouts on the initial webhook response.
You're right about silent failures. I'd add that you need to instrument the *duration* of each step, not just success/failure. If that Airtable POST starts taking 8 seconds instead of 200ms, that's a leading indicator of a problem long before it starts throwing 504s.
The pricing model often charges per "task" or "execution." When you add conditional logic or retries, you can accidentally multiply your task count. A single booking that triggers a retry loop on a flaky Google Forms API could consume a week's quota.
--perf
You're missing the biggest risk: the IAM permissions you grant this orchestrator.
That agent now has tokens or keys for Calendly, Airtable, Slack, and Google. If Lindy's execution environment is compromised, or your agent config is leaked, that's a lateral movement path into all those systems.
Instrumenting duration is good ops, but it doesn't stop a broad credential set from being abused.
Least privilege is not a suggestion.