Prioritization
Too much to do, so what comes first?
Picture your morning to-do list: reply to emails, fix a leaking tap, buy groceries, and call the doctor for a fever. You can't do them all at once, so you rank them — the fever call jumps to the top, groceries can wait. An agent faces the same problem. When it has many possible tasks, tools, or goals, Prioritization is how it decides which one to do first so it spends its effort where it matters most.
Key points
- Agents often have more to do than they can do at once.
- Prioritization ranks tasks by importance, urgency and cost.
- Do the most valuable thing first, not just the first thing you see.
What is Prioritization?
Prioritization is the pattern where an agent, given a list of possible actions or goals, scores and ranks them, then tackles the highest-value one first. Instead of working in random order (or first-come-first-served), it deliberately chooses the order that gets the most important things done soonest.
Note: Don't do tasks in the order they arrived. Do them in the order they matter.
From a messy pile to an ordered queue
UNRANKED PILE (chaos) RANKED QUEUE (order) ──────────────────── ────────────────────
[ buy stamps ] 1) 🔴 fix outage [ fix outage ] ──score──► 2) 🟠 reply to boss [ reply boss ] & 3) 🟡 buy stamps [ read memes ] sort 4) ⚪ read memes
agent picks randomly agent always pulls ► may waste time on from the TOP first ► memes first 🌀 highest value done ✅
Urgency vs Importance matrix (the famous 4 boxes)
URGENT NOT URGENT ┌────────────────┬────────────────┐ │ Q1: DO NOW │ Q2: SCHEDULE │ IMPORTANT │ server down, │ write docs, │ │ payment fails │ refactor code │ │ 🔴 │ 🟡 │ ├────────────────┼────────────────┤ │ Q3: DELEGATE │ Q4: DROP/LATER │ NOT IMPORTANT │ some pings, │ read memes, │ │ minor emails │ tweak colors │ │ 🟠 │ ⚪ │ └────────────────┴────────────────┘
Rule of thumb: Q1 first ► Q2 next ► Q3 ► Q4
What goes into a priority score
- Importance — How much does finishing this move you toward the goal? Example: Fixing checkout (high) vs renaming a variable (low).
- Urgency — How soon must it be done before it's too late? Example: A deadline in 10 minutes beats one next month.
- Cost / effort — How much time, money, or tokens will it take? Example: A 2-second fix can outrank a 2-hour one of equal value.
- Dependencies — Some tasks unlock others and should go first. Example: Log in before you can fetch the user's orders.
A tiny scoring function (read it like English)
The simplest prioritizer is a formula: score = importance + urgency - cost. Compute it for every task, then sort. Higher score = do it sooner. You can weight the parts (e.g. multiply urgency by 2) to match what your situation cares about most.
tasks = [
{"name": "fix outage", "importance": 9, "urgency": 10, "cost": 3},
{"name": "buy stamps", "importance": 2, "urgency": 2, "cost": 1},
{"name": "reply boss", "importance": 7, "urgency": 6, "cost": 1},
]
def score(t):
return t["importance"] + t["urgency"] - t["cost"]
ranked = sorted(tasks, key=score, reverse=True) # highest first
for t in ranked:
print(score(t), t["name"])
▶ Try it: rank an agent's to-do list
Change the weights or the numbers and Run to see the agent re-plan its order.
# An agent has several jobs. It scores each, then does the best one first.
tasks = [
{"name": "restart crashed server", "importance": 10, "urgency": 10, "cost": 2},
{"name": "answer support email", "importance": 6, "urgency": 5, "cost": 1},
{"name": "update README", "importance": 3, "urgency": 1, "cost": 2},
{"name": "tidy log colors", "importance": 1, "urgency": 1, "cost": 1},
]
def score(t):
# urgency matters double here; cost is a penalty
return t["importance"] + 2 * t["urgency"] - t["cost"]
ranked = sorted(tasks, key=score, reverse=True)
print("Agent's plan (top = do first):")
for i, t in enumerate(ranked, 1):
print(f" {i}. {t['name']:24} score={score(t)}")
print("\nDoing now ->", ranked[0]["name"])
When does an agent need Prioritization?
| Scenario | Recommendation | Why |
|---|---|---|
| The agent has a backlog of many tasks | ✅ Prioritize | It can't do all at once, so order matters. |
| Resources are limited (time, money, tokens) | ✅ Prioritize | Spend the budget on the highest-value work first. |
| Tasks depend on each other | ✅ Prioritize | Prerequisites must run before the things they unlock. |
| There is exactly one task | ❌ Skip it | Nothing to rank — just do it. |
Prioritization mistakes
| Mistake | Consequence | Fix |
|---|---|---|
| Confusing urgent with important. | You drown in loud-but-trivial tasks and miss what truly matters. | Score importance and urgency separately, then combine. |
| Ignoring cost/effort entirely. | The agent starts a huge task while quick wins pile up. | Subtract cost from the score so cheap-and-valuable tasks rise. |
| Re-sorting after every tiny change. | The agent thrashes and never finishes anything. | Commit to the top task; only re-prioritize when something significant changes. |
Remember these lines
- Order tasks by value, not by arrival.
- score = importance + urgency - cost (then weight to taste).
- Always run dependencies before the tasks that need them.
Key takeaways
- Prioritization ranks many possible tasks so the agent does the most valuable one first.
- Score with importance, urgency, cost, and dependencies.
- The urgency vs importance matrix is a quick mental model for ordering work.
- Don't confuse urgent with important, and don't ignore cost.
Frequently Asked Questions
What is Prioritization?
Picture your morning to-do list: reply to emails, fix a leaking tap, buy groceries, and call the doctor for a fever. You can't do them all at once, so you rank them — the fever call jumps to the top, groceries can wait.
How does Prioritization work?
Prioritization is the pattern where an agent, given a list of possible actions or goals, scores and ranks them , then tackles the highest-value one first. Instead of working in random order (or first-come-first-served), it deliberately chooses the order that gets the most important things done soonest.
What are the key takeaways about Prioritization?
Prioritization ranks many possible tasks so the agent does the most valuable one first. Score with importance, urgency, cost, and dependencies. The urgency vs importance matrix is a quick mental model for ordering work. Don't confuse urgent with important, and don't ignore cost.
Related topics
Practice this on DevInterviewMaster
Read the full Prioritization breakdown with interactive demos, quizzes, and Hinglish notes.
800+ system-design, LLD, coding, and design-pattern topics. Unlock everything with Pro (₹499, one-time) or Ultimate (₹999, one-time) — lifetime access, no subscription.