Caching and result reuse in Agentic AI - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Why is caching important in agentic AI workflows?

Imagine you have an AI agent that performs multiple steps to answer a question. Why would caching intermediate results help?

A It makes the agent forget previous steps faster.

B It increases the randomness of the agent's answers.

C It reduces repeated calculations, saving time and resources.

D It forces the agent to always start from scratch.

💡 Hint

Think about how computers avoid doing the same work twice.

Caching stores results of previous computations so the agent can reuse them instead of recalculating, which speeds up processing and saves resources.

❓ Predict Output

intermediate

What is the output of this caching example?

Consider this Python code simulating caching in an AI agent:

cache = {}
def expensive_step(x):
    if x in cache:
        return cache[x]
    result = x * x  # Simulate expensive calculation
    cache[x] = result
    return result

print(expensive_step(3))
print(expensive_step(3))
print(expensive_step(4))

What will be printed?

A

9
9
9

B

9
None
16

C

3
3
4

D

9
9
16

💡 Hint

Check if the function uses cached results for repeated inputs.

The first call calculates 3*3=9 and caches it. The second call returns cached 9. The third call calculates 4*4=16.

❓ Model Choice

advanced

Which caching strategy best suits agentic AI with changing inputs?

You have an AI agent that processes user queries with slight variations. Which caching method helps reuse results without returning outdated answers?

A Cache results with approximate matching and expiration time.

B Do not use caching at all.

C Cache all results indefinitely without checking input.

D Cache results with exact input keys only.

💡 Hint

Think about balancing reuse and freshness of results.

Approximate matching with expiration allows reuse of similar inputs while avoiding stale data, ideal for dynamic queries.

❓ Metrics

advanced

How does caching affect AI agent performance metrics?

An AI agent uses caching to store intermediate results. Which metric is most directly improved by caching?

A Model accuracy on test data

B Inference latency (time to get a result)

C Training loss during model training

D Number of model parameters

💡 Hint

Consider what caching saves in repeated computations.

Caching reduces the time needed to produce results by avoiding repeated calculations, lowering inference latency.

🔧 Debug

expert

Why does this caching code cause incorrect reuse?

Review this Python snippet for caching AI agent results:

cache = {}
def process(input_data):
    if 'result' in cache:
        return cache['result']
    result = input_data + 1
    cache['result'] = result
    return result

print(process(5))
print(process(10))

What is the problem with this caching approach?

A It caches only one result regardless of input, causing wrong reuse.

B It does not cache any results, so no reuse happens.

C It raises a KeyError because 'result' key is missing.

D It caches results correctly for each input.

💡 Hint

Check how the cache key is used for different inputs.

The cache uses a fixed key 'result' for all inputs, so the second call reuses the first result incorrectly.