AdlyseAdlyseDocs
Welcome
Getting Started
QuickstartOnboarding Wizard
Essentials
DashboardAd AccountsExplorerReportsSearch TermsDecision Log
AI Studio
Adlyse AISkillsGuidelinesSchedules
Admin
OrganizationTeam
Integrations
OverviewMCP Servers →API Keys
Legacy Docs
OverviewQuickstartCookbookWorkflowVariablesKnowledge BaseAPI Keys (legacy)NodesSelf HostingSelf-Host Auth
  1. Ai Studio

Guidelines

Guidelines are short markdown documents that shape how Adlyse’s AI behaves for your organization. They’re the canonical place for brand voice, industry context, platform-specific playbooks, and any rules the assistant must follow.

The four system slugs#

Every org has four system guidelines:

SlugPurpose
adlyseOverarching assistant instructions — tone, caution level, escalation rules
brandBrand voice, audience, product, taboos
googleGoogle Ads-specific playbook (platform-gated: only injected for Google accounts)
metaMeta Ads-specific playbook (platform-gated)

You can also create org-specific slugs for internal rules that don’t fit a system slug.

When guidelines are injected#

  • adlyse and brand are always included in the assistant’s system prompt.
  • google and meta are conditionally injected based on which ad accounts the current thread has access to.
  • All four are readable through adlyse-insights.list_guidelines / get_guideline so external agents can load the same context.

Editing#

From AI Studio → Guidelines, pick a slug and edit the markdown. Saves go through a short review lane: new content lives as a draft until an admin marks it production. Old versions remain readable so you can roll back or inspect what was live when a past decision was made.

The assistant itself can propose draft updates through the suggest_guideline_draft tool — when it notices a missing rule, a consistently useful pattern, or outdated instruction over the course of a conversation. Drafts appear in the same review lane; a human always promotes them.

Seeded at onboarding#

When you first sign up and complete the onboarding wizard, Adlyse runs a background brand-research job that scrapes your website, analyzes competitor URLs you provide, and produces a first-pass brand guideline. You can keep, edit, or discard it.

←Previous
Skills
Next→
Schedules