Unified API — Baselines and variants share the same document type model and support the same operations. Three tasks become more specific: analyzing content structures via referrers, accessing content items via observed properties, and navigating between sites in a multi-site setup. For these, the API adds baseline-only accessors (for example, getBaselineReferrersWithDescriptor, getBaselineContentsWithValue, and the getAllVariants variants of the ContentSiteAspect methods), and the Content.isVariant / Content.isBaseline flags. See the Unified API Developer Manual for details.

CAE — The CAE uses a client-side JavaScript library to replace baseline fragments with variant fragments dynamically, using metadata div elements and the p13n-dynamic-include brick. The Content Application Engine Developer Manual describes the extension points (BodyDetector, ContentVariantProcessor, VariantConditionPredicate), the programming model, and the FreeMarker library.

Headless Server — Each content object exposes a contentVariants field (and contentVariantsAndBaseline) that lists the possible variants in descending priority order, together with their conditionStrings and parsed conditions. The Headless Server Developer Manual describes the variant-selection algorithm and example GraphQL queries.

Search and Feeder — The Content Feeder indexes variants like baseline content and adds variant metadata (for example, variantBase, variantSegments, variantAbTestCase) to Solr. The CAE Feeder does not process variants, so a website search finds baseline content only.

Studio, tools and configuration — Studio maintains the variant folder location and enforces the variant meta-model through validators. The command line tools handle baselines and variants uniformly (see the variant marker in cm serverexport/cm serverimport), and cm cleanvariants removes left-over variants of deleted or destroyed baselines.