Uncertainty Planning
Master uncertainty planning to build robust agents that thrive in unpredictable environments
Your Progress
0 / 5 completedBuilding Backup Plans
Contingency planning means preparing alternative paths before you need them. Instead of hoping your plan works, you explicitly design what to do when it doesn't. This isn't pessimism—it's realism.
Types of Contingencies
Backup Plans (Plan B, C, D...)
Alternative approaches you can switch to if the primary plan fails. Same goal, different method.
Example: Travel Planning
Plan A:Take 9am flight (fastest)
Plan B:Take 2pm flight if 9am sold out
Plan C:Drive 5 hours if flights canceled
Plan D:Attend meeting remotely via video
When to Use
- • Mission-critical tasks
- • High-uncertainty situations
- • Time-sensitive goals
How Many Backups?
More uncertainty = more backups. Critical tasks might need 3-4 alternatives.
Interactive: Contingency Depth Calculator
Adjust risk level to see recommended contingency depth
50%
Low RiskHigh Risk
Contingency NeedImportant
Risk Meter
Recommendation:
Prepare 2-3 backup options - moderate chance of needing them
Plan A
Primary
Plan B
Backup
Plan C+
Multi-tier
Conditional Planning (If-Then-Else Logic)
Structure your plans as decision trees: "If X happens, do Y; else do Z."
if user_confirms_order():
process_payment()
send_confirmation_email()
elif payment_fails():
retry_with_backup_method()
notify_user_of_issue()
elif timeout():
save_cart_for_later()
send_reminder_email()
else:
log_error()
alert_support_team()
✅Good Contingency Planning
- •Specific triggers: "If API returns 429..."
- •Clear actions: "Then wait 60s and retry"
- •Multiple layers: Plans A, B, C defined
- •Realistic: Actually executable alternatives
❌Poor Contingency Planning
- •Vague: "If something goes wrong..."
- •Wishful: "Then just make it work somehow"
- •Single layer: Only one backup option
- •Unrealistic: Requires unavailable resources