This page consolidates the four working documents on the proposed NOTHS ↔ Similar.ai integration: canonical URL shape, system boundaries, catalog data we could send, and how generated shop content could be delivered.

Overview

This document describes possible ways NOTHS could connect its catalog and shop experience to Similar.ai. It covers the canonical PLP URL shape, system boundaries and source of truth, what catalog data we could send, how generated shop content could be delivered, and the workflow for approvals and publishing.

Canonical URLs (taxonomy-based PLP structure)

Unlike the existing category-based PLP URL structure, the planned taxonomy-based PLP URL structure would encode everything needed to query Vertex. ContentStack would supply optional editorial content only (for example headings and intro copy). As long as the URL is valid, the PLP would show products returned by Vertex — the CMS would not be required for the page to resolve with product results.

Base structure

The URL has two sections, separated by a literal dash (-) only when filters are present:

  1. Taxonomy path: {taxonomy}/{level1}/[{level2}/[{level3}/[{level4}/]]]
  2. Filters (optional): after /-, zero or more {attribute}/{value} pairs.
Segment Requirement Description
taxonomy Mandatory Hierarchy root: shop or gift.
level1 Mandatory Top-level category (for example jewellery).
level2level4 Optional Sub-categories; include the segments needed when skipping levels to reach a deeper category.
- Conditional Separator between the taxonomy path and filters. Required only if filters are present.
filter Optional Pairs of {attribute}/{value}.

Formal grammar (regex-style)

/{taxonomy}/{l1}/[{l2}/[{l3}/[{l4}/]]][/-/{attr}/{val}(/{attr}/{val})*]

Similar.ai-facing payloads that reference a path (for example async pushes or on-demand API requests) would need to stay consistent with this URL scheme wherever we treat those paths as canonical PLP URLs.

System boundaries

At a high level, responsibilities would split as follows: