Just spent three days trying to get a Pipedrive "project" to behave like a proper support queue. Their "custom fields" and "activities" feel like they're designed to *look* flexible.
The trap is believing you can mold it into something it's not. You start adding custom fields, then you need automations to glue the logic together, then you're writing API middleware to fill the gaps. Suddenly, your "simple CRM" is a distributed system held together with webhooks and hope.
```yaml
# Example: Their API response for a "custom" deal
deal:
id: 123,
title: "Seems fine",
custom_fields: {
"a1b2c3": "Actual value" # Good luck remembering this mapping
},
# Where's the pipeline stage history? Oh right, separate endpoint.
```
You end up building your own state machine on top of theirs. For the same effort, you could've just run a lightweight OSS CRM in a k8s pod with actual schema control. Their model is rigid—pipelines, deals, activities, period. Trying to make it a project manager, a support desk, or an inventory system is where the pain begins 😅
Anyone else found that after a certain complexity threshold, Pipedrive starts fighting you? What was your breaking point?