A minimal Planner → Creator → Critic agent loop to test whether YouTube content operations can auto-produce daily. $29/mo, 54% margin (pre-optimization).
If RegWatch validated “narrow vertical + high margin,” OPC Lite validated something else: whether the smallest automation loop for a content-creator OPC actually holds up.
We deliberately kept v1 small — serving only content creators (YouTube content operations), running just the shortest agent loop:
In the MVP phase we asked one question: can a creator get a usable batch of output every day without doing it themselves?
This loop rolls daily, compressing “hours a creator would spend” into “a few minutes of review.”
Many demos only show pretty numbers. We publish 54% because it reveals the real cost curve:
The MVP had no model routing — everything ran on Sonnet, so per-OPC model cost was about $17.75/month.
Our optimization roadmap is clear:
| Stage | Model cost / OPC / mo | Lever |
|---|---|---|
| MVP (pre-routing) | $17.75 | All Sonnet |
| After routing | ~$6 | L1–L4 tiers + Prompt Cache |
| After private LoRA | ~$3 | 40% of traffic to private models |
| At scale (target) | < $2 | Routing + cache + private models |
So 54% isn’t the ceiling — it’s the starting point before optimization. For the same OPC, cutting model cost from $17.75 to under $2 rewrites the entire margin structure.
Want to try your own Content Creator OPC? Let’s talk.