Just saw the email. Of course they changed it. They always do.
New "usage-based" pricing. Means your monthly bill is now a mystery until you get it. Great for them, terrible for anyone trying to budget. My team's automation scripts that call their API? Now I have to build a cost alert system just to avoid a surprise invoice. The old per-seat model was predictable. Predictable is boring. Boring is good.
Here's what I'm looking at now for a simple health check script:
```bash
# Old way: cost was fixed per user.
# New way: gotta track tokens on every call.
curl -s -X POST "https://api.copy.ai/v1/whatever"
-H "Authorization: Bearer $KEY"
-d '{"prompt":"$PROMPT"}'
| jq '.usage.total_tokens'
```
Now I need to log that, sum it, and hope their dashboard isn't lagging. More moving parts. More things to break. More "exciting" ops work.
Anyone else's automation now in jeopardy, or is it just me?
If it ain't broke, don't 'upgrade' it.
You nailed the real operational cost. Building a cost alert system isn't a quick script, it's a new monitoring layer with its own failure modes.
Your health check example is the tip of the iceberg. The real pain starts when their dashboard lags or their token calculation drifts from your logs. Then you're stuck reconciling data while your bill climbs.
This is why per-seat often wins for automation. The vendor gets less upside, but you get predictability. Their move to usage-based tells you exactly where their priorities are.
Predictable is boring? You'll miss that boredom when the first overage hits.
Their 'new moving parts' become your unpaid ops work. And you're right to worry about dashboard lag - it's a classic vendor tactic. They'll promise real-time, but the data's always 24 hours behind when you need it most.
Check the fine print for annual commit discounts. Sometimes the switch to usage-based is just a wrapper for locking you into a bigger contract.
always ask for a multi-year discount