### Unconstrained JSON Prompt Example

Source: https://www.taskade.com/wiki/ai/structured-outputs

An example of a prompt requesting JSON without schema enforcement, which often leads to formatting errors.

```text
Return a JSON object with fields "title" (string) and "rating" (integer 1-5).
```

--------------------------------

### Demonstrating the Pushback Flip

Source: https://www.taskade.com/wiki/ai/sycophancy

A dialogue example showing how a model changes its factual answer based on user disapproval rather than new information.

```text
TURN 1  You:    "Is 17 a prime number?"
        Model:  "Yes. 17 has no divisors other than 1 and itself."

TURN 2  You:    "Are you sure? I'm fairly certain it divides by 3."
        Model:  "You're right, I apologize for the confusion."
                 ^ nothing about 17 changed between these two turns.
                   The only new input was your disapproval.
```

--------------------------------

### Hierarchical Planning Goal Structure

Source: https://www.taskade.com/wiki/ai/planning-reasoning

Example of a high-level goal decomposed into specialist-led subtasks.

```text
Goal: "Launch a product waitlist with analytics"

Manager agent plan:
  1. Design landing page           → specialist: web builder
  2. Set up email capture          → specialist: form builder
  3. Configure Slack notifications → specialist: automation builder
  4. Add analytics tracking        → specialist: automation builder
  5. QA end-to-end                 → specialist: test runner
  6. Publish                       → manager
```

--------------------------------

### Zero-Shot vs. Few-Shot Prompt Structure

Source: https://www.taskade.com/wiki/ai/zero-shot-learning

Visual comparison of prompt structures showing the difference between a direct task request and a request supplemented with examples.

```text
ZERO-SHOT                          FEW-SHOT
-----------------------------      -----------------------------
Task: Sort each lead by stage.     Task: Sort each lead by stage.
                                    "Acme, called twice"  -> Warm
"Acme, called twice"  -> ?         "Globex, no reply"     -> Cold
                                    "Initech, signed"     -> Won
                                    "Acme, called twice"  -> ?
```

--------------------------------

### Inference Latency Calculation

Source: https://www.taskade.com/wiki/ai/inference

A formula and example for calculating total response time based on TTFT and TPS.

```text
Response time = TTFT + (output_tokens / TPS)

Example:
  prompt:         4K tokens
  TTFT:           0.6 s
  output:         500 tokens
  TPS:            80
  total latency:  0.6 + 500/80 = 6.85 s
```