01
Low
Direct execution with minimal deliberation.
Edits, summaries, extraction, formatting Independent AI task router
Stop paying your AI to second-guess itself.
More thinking isn't always better thinking. Paste a task and get the right effort level before you burn the time and tokens.
Route a task01 / THE ADVISOR
What are you asking AI to do?
We score task complexity, uncertainty, stakes, and duration—then recommend the smallest thinking budget that fits.
Try an example:
YOUR RECOMMENDATION
Ready when you are.
Add a task and three quick signals. No login. No prompt data leaves this page.02 / THE SCALE
Four levels. One useful default.
Start at medium. Move up only when the task gives you a reason—not because the button exists.
02DEFAULT
Medium
Enough reasoning for most real work.
Writing, coding, planning, analysis03
High
More exploration for ambiguity and tradeoffs.
Architecture, research, debugging, strategy04
Max
Sustained reflection across a long-running task.
Autonomous builds, migrations, complex investigations03 / THE PRINCIPLE
The best setting is not the highest setting. It's the lowest one that reliably works.
01
Effort is a budget. It controls how long the model searches, reflects, and retries—not its raw intelligence.
02
Simple work wants momentum. Excess reasoning can invent complexity and second-guess a correct first answer.
03
Harnesses matter. Tools, context, and checkpoints often improve the result more than another notch of effort.
04 / BUSINESS MODEL
Free to decide. Paid to automate.
The public advisor earns trust. Pro routes tasks inside your daily tools. Teams pay for policies, budgets, and proof of savings.
$0
Forever
- Unlimited task checks
- Prompt starters
- Private, in-browser scoring
EARLY ACCESS
PRO$9
Per month
- Browser + IDE advisor
- Automatic model routing
- Usage and savings history
$49
Per workspace / month
- Shared routing policies
- Cost and latency dashboard
- Admin controls + API