If you’ve used Claude’s custom skills, you’ll get the idea. AI Skills brings the same concept to Holistics AI: instead of cramming everything into a single context.aml file, Analysts can define multiple skill definitions right in AML, each scoped to specific teams with its own instructions, data references, and access controls.
Here’s what you’ll be able to do:
- Create domain-specific skills: a
finance_skilland anecommerce_skillcan coexist, each with its own instructions, terminology, and dataset references - Control access per user: use user attribute expressions to restrict skills by department, role, or any attribute (e.g.,
disabled: user.department != 'Finance') - Set activation modes: skills can auto-activate when relevant, require manual invocation via slash commands, or stay hidden as background context
- Reference contexts dynamically: inject datasets, contexts, or files on demand, keeping the AI’s context lean until content is actually needed
- Import from existing docs: upload markdown skill files and convert them into AML definitions
End-users interact with skills through slash commands, UI toggles, or just by chatting. The AI picks the right skill automatically.
Skills are defined as an AML type, so they live in your data model alongside everything else:
Skill finance_analysis {
label: "Finance Analysis"
description: "Specialized AI skill for finance team queries"
disabled: user.department != 'Finance'
content: |-
You are a finance analyst assistant.
Use @Dataset:finance_metrics for all queries.
}
AI Skills is coming soon. Share your thoughts and use cases below!