daBongo LMS AI Training Courses

Lesson Content

Layered context building.

Each context layer narrows to what is actually relevant:

Layer 1 – Role: "I am a nurse" Layer 2 – Setting: "working in a community health clinic" Layer 3 – Situation: "who needs to explain a new medication regimen to patients with low health literacy" Layer 4 – Constraint: "without using medical jargon" Layer 5 – Goal: "so they can follow it accurately at home"

Combined, these layers transform a generic medical communication question into a response targeted to the actual situation.

Time and recency constraints.

For any topic where currency matters: "Current best practices" / "most recent guidance" / "as of 2026" / "what has changed in the past two years"

Without time constraints, Meta AI may draw on older patterns in its training.

Tone and audience constraints.

"Professional but approachable – not corporate or stiff" "Plain language – assume no professional background in this area" "Direct and brief – this person is busy and reads fast"

Tone and audience constraints produce output that actually sounds right for the receiver.

Negative constraints – what to exclude.

"Do not suggest anything requiring a gym membership" "Exclude technical jargon – practical language only" "Do not provide generic advice that would apply to anyone – I need this specific to my situation"

Negative constraints are efficient – they eliminate irrelevant content categories without requiring you to specify every acceptable alternative.

Structured output formats.

For decisions: "Present as: recommendation first, then three supporting reasons, then the main counterargument and my response to it."

For step-by-step guidance: "Numbered steps. Each step: what to do, what to avoid, and an example."

For comparison: "Side-by-side comparison table: rows are criteria I care about, columns are each option."

The one-constraint-at-a-time refinement.

When a response is not quite right, change one thing at a time:

• Too formal – "Make this warmer and more conversational"
• Too long – "Shorten to one paragraph"
• Wrong focus – "Remove the section on X, expand the section on Y"

Changing multiple things at once makes it harder to identify what improved the response.