Custom Context
Custom context lets admins define a shared knowledge base for every AI interaction across your organization — business background, internal terminology, data conventions, and response preferences that your semantic layer can't express on its own.
Where to find it
Custom context lives in context.aml, accessible from:
- Development > Project Explorer >
settings>ai - Tenant name > Organization Settings > AI Settings
What to put in it
The file accepts freeform Markdown under const ai_context = @md. Some useful content includes:
- Business background — Who is your company, what do you sell, who are your customers?
- Operational knowledge — Business rules and definitions not captured in your data models. If your company defines "churn" differently from the industry standard, document it here.
- Response preferences — How should AI communicate? Preferred language, tone, or answer format.
Examples
Add business background
const ai_context = @md
## Business background
Adventure Works sells high-quality road bikes, mountain bikes,
and cycling accessories including bike racks, stands, tires, and tubes.
;;
Add response preferences
const ai_context = @md
## Response preferences
- Be concise, professional, and objective
- Always include 1–2 key insights and suggest next steps for deeper analysis
;;
Customize context by team
Context can be dynamic. Reference user attributes to give different teams responses tailored to their workflows and datasets.
const product_context = '''
Preferred dataset: ${products_dataset}
Typical analyses are:
- Which products are sold the most?
- Which has the highest margin?
'''
const customer_context = '''
Preferred dataset: ${customers_dataset}
Typical analyses are:
- Where are my customers most concentrated?
- What do the highest-paying customers look like?
'''
const ai_context = @md
${
if (H.current_user.team == 'Business Analyst') {
product_context
} else if (H.current_user.team == 'Growth & Marketing') {
customer_context
}
}
;;