Fetch Framework — Live Use Cases

Quick Reference

The Formula

Fetch = Chirp × |DRIFT| × Confidence

Where:
  Chirp      = Signal strength (0-100)
  DRIFT      = Methodology − Performance (the gap)
  Confidence = min(Perch, Wake) / 100

Decision Thresholds

Fetch Score	Action	Meaning
> 1000	Execute	High confidence, clear signal, real gap
500 - 1000	Confirm	Ask before proceeding
100 - 500	Queue	Log for later review
< 100	Wait	Insufficient conditions

The Three Questions

Dimension	Question	Low Score Means
Chirp	"Should I act?"	No urgency
DRIFT	"How much?"	No gap to close
Confidence	"Can I?"	Not ready

---

Use Case 1: Browser Automation Agent

Scenario

User command: "Click the Submit button on this form"

Dimension Scores

Dimension	Score	Reasoning
Chirp	90	Clear, direct command. Action verb "Click" detected.
Perch	75	Found button with id="submit-btn". 75% selector confidence (other similar buttons exist).
Wake	85	Context from previous steps: user filled form fields.
DRIFT	60	Gap between current state (form filled) and goal (form submitted).

Calculation

Confidence = min(75, 85) / 100 = 0.75

Fetch = 90 × 60 × 0.75
Fetch = 4,050

Decision

Fetch	Threshold	Action
4,050	> 1000	✅ Execute — Click the button

Post-Action

After execution:

• New DRIFT = 5 (form submitted, near goal)
• Re-run Fetch for next action or complete

---

Use Case 2: Content Publishing Decision

Scenario

You have a video ready. Should you publish now or wait?

Dimension Scores

Dimension	Score	Reasoning
Chirp	45	Moderate trending signal. Topic relevant but not urgent.
Perch	80	Strong structure. Good thumbnail, title, description ready.
Wake	70	Fits channel history. Audience expects this content type.
DRIFT	35	Small gap. Content is close to target methodology.

Calculation

Confidence = min(80, 70) / 100 = 0.70

Fetch = 45 × 35 × 0.70
Fetch = 1,102

Decision

Fetch	Threshold	Action
1,102	> 1000	✅ Execute — Publish now

Alternative Scenario: Low Signal

Same content, but Chirp = 20 (no trending relevance):

Fetch = 20 × 35 × 0.70
Fetch = 490

Fetch	Threshold	Action
490	500 - 1000	⚠️ Confirm — Consider waiting for better timing

---

Use Case 3: Sales Call Prioritization (CVA/PSSR)

Scenario

AI assistant preparing for customer call. Should it generate talking points?

Dimension Scores

Dimension	Score	Reasoning
Chirp	95	Meeting in 5 minutes. High urgency.
Perch	60	Customer data available but scattered across 7 systems.
Wake	40	Limited call history. New rep, unfamiliar with customer.
DRIFT	80	Large gap between current prep (nothing) and goal (ready).

Calculation

Confidence = min(60, 40) / 100 = 0.40

Fetch = 95 × 80 × 0.40
Fetch = 3,040

Decision

Fetch	Threshold	Action
3,040	> 1000	✅ Execute — Generate summary now

Insight

High urgency (Chirp = 95) and large gap (DRIFT = 80) overcome low confidence (0.40). The system acts because not acting has higher cost.

Low Urgency Variant

Same scenario, but meeting is tomorrow (Chirp = 30):

Fetch = 30 × 80 × 0.40
Fetch = 960

Fetch	Threshold	Action
960	500 - 1000	⚠️ Confirm — Suggest gathering more context first

---

Use Case 4: Trading Decision

Scenario

Signal detected for potential trade entry.

Dimension Scores

Dimension	Score	Reasoning
Chirp	70	Breakout signal. Above average but not extreme.
Perch	85	Clear support/resistance levels. Good R:R structure.
Wake	90	Pattern matches historical setups. Strong backtest.
DRIFT	25	Small gap. Price near entry zone already.

Calculation

Confidence = min(85, 90) / 100 = 0.85

Fetch = 70 × 25 × 0.85
Fetch = 1,487

Decision

Fetch	Threshold	Action
1,487	> 1000	✅ Execute — Enter trade

Failed Setup

Same signal, but no historical pattern match (Wake = 15):

Confidence = min(85, 15) / 100 = 0.15

Fetch = 70 × 25 × 0.15
Fetch = 262

Fetch	Threshold	Action
262	100 - 500	📋 Queue — Log for review, don't trade

Insight

High Chirp and Perch aren't enough. Without Wake (memory/pattern), confidence collapses. The system won't act on signals without historical validation.

---

Use Case 5: Engineer Hours Estimation

Scenario

New project came in. Should AI generate an estimate or defer to domain expert?

Dimension Scores

Dimension	Score	Reasoning
Chirp	50	Moderate urgency. Quote needed this week.
Perch	45	Partial data. Model type known, scope unclear.
Wake	30	Few similar projects in history. Sparse training data.
DRIFT	70	Large gap between current (no estimate) and goal (accurate hours).

Calculation

Confidence = min(45, 30) / 100 = 0.30

Fetch = 50 × 70 × 0.30
Fetch = 1,050

Decision

Fetch	Threshold	Action
1,050	> 1000	✅ Execute — But...

The Nuance

Fetch score crosses threshold, but confidence is only 0.30. System should:

1. Generate estimate (Fetch > 1000)
2. Flag low confidence visibly
3. Recommend domain expert review

Enhanced Threshold Logic

if (Fetch > 1000 && Confidence > 0.60) {
  execute()           // Full automation
} else if (Fetch > 1000 && Confidence <= 0.60) {
  executeWithFlag()   // Automate but flag for review
} else if (Fetch > 500) {
  confirm()           // Human decides
} else {
  wait()              // Don't act
}

---

Use Case 6: Document Generation

Scenario

User asks: "Create a presentation for the CVA meeting"

Dimension Scores

Dimension	Score	Reasoning
Chirp	85	Direct request. Clear intent.
Perch	70	Template available. Structure known.
Wake	75	Previous presentations in history. Style established.
DRIFT	90	Large gap. No presentation exists yet.

Calculation

Confidence = min(70, 75) / 100 = 0.70

Fetch = 85 × 90 × 0.70
Fetch = 5,355

Decision

Fetch	Threshold	Action
5,355	> 1000	✅ Execute — Generate presentation

---

Use Case 7: Email Response

Scenario

AI assistant monitoring inbox. New email arrives. Should it draft a response?

Dimension Scores (Important Email)

Dimension	Score	Reasoning
Chirp	80	From VP. Marked urgent.
Perch	65	Clear question asked. Response structure obvious.
Wake	85	Long thread history. Context rich.
DRIFT	50	Moderate gap. Some info needed for complete response.

Calculation

Confidence = min(65, 85) / 100 = 0.65

Fetch = 80 × 50 × 0.65
Fetch = 2,600

Decision: Execute — Draft response

Dimension Scores (Newsletter)

Dimension	Score	Reasoning
Chirp	5	Marketing email. No urgency.
Perch	20	No clear action required.
Wake	10	No prior engagement with sender.
DRIFT	15	Tiny gap. Archive is fine.

Calculation

Confidence = min(20, 10) / 100 = 0.10

Fetch = 5 × 15 × 0.10
Fetch = 7.5

Decision: Wait — No action needed

---

Use Case 8: Code Refactoring

Scenario

AI suggests refactoring a function. Should it auto-apply?

Dimension Scores (Safe Refactor)

Dimension	Score	Reasoning
Chirp	40	Suggestion, not urgent request.
Perch	90	Clear function boundaries. Well-structured code.
Wake	95	Strong test coverage. Safe to change.
DRIFT	30	Small improvement. Code works but could be cleaner.

Calculation

Confidence = min(90, 95) / 100 = 0.90

Fetch = 40 × 30 × 0.90
Fetch = 1,080

Decision

Fetch	Threshold	Action
1,080	> 1000	✅ Execute — Apply refactor

Dimension Scores (Risky Refactor)

Dimension	Score	Reasoning
Chirp	40	Same suggestion.
Perch	50	Tangled dependencies.
Wake	25	No tests. Unknown side effects.
DRIFT	30	Same improvement potential.

Calculation

Confidence = min(50, 25) / 100 = 0.25

Fetch = 40 × 30 × 0.25
Fetch = 300

Decision

Fetch	Threshold	Action
300	100 - 500	📋 Queue — Log suggestion, don't auto-apply

---

Pattern Summary

What Drives High Fetch?

Factor	Effect
High Chirp	Urgency pushes action
High DRIFT	Large gap demands closing
High Confidence	Perch + Wake alignment enables action

What Blocks Fetch?

Factor	Effect
Chirp = 0	No signal → No action
DRIFT = 0	No gap → No need to act
Confidence → 0	Either Perch or Wake near zero → Can't act safely

The Multiplication Effect

Because Fetch is multiplicative, one weak dimension kills the score:

Chirp	DRIFT	Confidence	Fetch
90	80	0.85	6,120 ✅
90	80	0.10	720 ⚠️
90	5	0.85	382 📋
10	80	0.85	680 ⚠️

---

Implementation Template

function calculateFetch(chirp, perch, wake, methodology, performance) {
  const drift = Math.abs(methodology - performance);
  const confidence = Math.min(perch, wake) / 100;
  const fetch = chirp * drift * confidence;
  
  return {
    fetch,
    drift,
    confidence,
    action: getAction(fetch, confidence)
  };
}

function getAction(fetch, confidence) {
  if (fetch > 1000 && confidence > 0.60) {
    return 'execute';
  } else if (fetch > 1000 && confidence <= 0.60) {
    return 'execute_with_review';
  } else if (fetch > 500) {
    return 'confirm';
  } else if (fetch > 100) {
    return 'queue';
  } else {
    return 'wait';
  }
}

---

Quick Decision Matrix

Urgency (Chirp)	Gap (DRIFT)	Ready (Confidence)	Action
High	Large	High	Execute
High	Large	Low	Execute + Review
High	Small	Any	Confirm
Low	Large	High	Confirm
Low	Large	Low	Queue
Low	Small	Any	Wait
Any	Zero	Any	Wait (no gap)

---

The Feedback Loop

┌─────────────────────────────────────┐
│  1. Measure dimensions              │
│     Chirp, Perch, Wake              │
│              ↓                      │
│  2. Calculate DRIFT                 │
│     Methodology − Performance       │
│              ↓                      │
│  3. Calculate Fetch                 │
│     Chirp × |DRIFT| × Confidence    │
│              ↓                      │
│  4. Decide action                   │
│     Execute / Confirm / Queue / Wait│
│              ↓                      │
│  5. Execute (if threshold met)      │
│              ↓                      │
│  6. Re-measure                      │
│     DRIFT should decrease           │
│              ↓                      │
│  7. Loop until DRIFT ≈ 0            │
└─────────────────────────────────────┘

---

Framework Integration

Framework	Role in Fetch
ChirpIQX	Provides urgency signal
PerchIQX	Provides structural confidence
WakeIQX	Provides historical confidence
DRIFT	Provides gap measurement
Fetch	Synthesizes all into action decision

---

"DRIFT sees the distance. Fetch closes it."