Research and draft first mapping Drupal AI vector database layer to Symfony AI Store
>>> [!note] Migrated issue
<!-- Drupal.org comment -->
<!-- Migrated from issue #3567005. -->
Reported by: [marcus_johansson](https://www.drupal.org/user/385947)
>>>
<p>[Tracker]<br>
<strong>Update Summary: </strong>[One-line status update for stakeholders]<br>
<strong>Short Description: </strong>Research and draft first mapping Drupal AI vector database layer to Symfony AI Store<br>
<strong>Check-in Date: </strong>MM/DD/YYYY<br>
[/Tracker]</p>
<h3 id="summary-problem-motivation">Problem/Motivation</h3>
<p>A core requirement for adopting Symfony AI is that Drupal AI’s vector database abstraction and integrations can be replaced or reimplemented using Symfony AI Store. Without this, we cannot switch to Symfony AI as the underlying framework.</p>
<p>Drupal AI’s vector database layer is a foundational piece for retrieval workflows and any RAG-related features, and it likely has multiple provider implementations and extension points that contrib modules may rely on. We need an explicit first mapping between Drupal AI’s current vector store responsibilities and Symfony AI Store’s APIs and capabilities, along with a rough initial idea for how implementation could work inside Drupal AI.</p>
<p>This research depends on:<br>
- The documented understanding of Drupal AI’s current capabilities (including vector DB features and extension points).<br>
- The structured document of open questions and unknowns identified during the exploratory phase.</p>
<h3 id="summary-proposed-resolution">Proposed resolution</h3>
<p>- Review the Drupal AI capabilities document and extract all vector database responsibilities, supported providers, and integration points.<br>
- Review the open questions/unknowns document and highlight which questions relate directly to vector databases, indexing, retrieval, and RAG workflows.<br>
- Analyze Symfony AI Store APIs and concepts and map them to Drupal AI’s current vector database layer:<br>
- Identify direct equivalents where Drupal can become a thin adapter.<br>
- Identify mismatches where Drupal AI needs glue code or alternate patterns.<br>
- Identify missing capabilities that would require upstream Symfony AI changes or Drupal-side workarounds.<br>
- Draft a first-pass “mapping” document that includes:<br>
- A side-by-side mapping of key concepts (Drupal AI vector responsibilities → Symfony AI Store equivalents).<br>
- A proposed Drupal-side architecture for integration (rough implementation idea), including:<br>
- Where Symfony AI Store would be instantiated and configured.<br>
- How Drupal configuration (provider selection, credentials, index configuration) maps to Symfony Store concepts.<br>
- How indexing and retrieval flows would be triggered from Drupal AI.<br>
- How existing Drupal AI APIs and plugin points could wrap Symfony Store without breaking contrib expectations.<br>
- How provider-specific concerns (schema/index differences, embedding models, distance functions) would be handled after the change.<br>
- Summarize feasibility and risks:<br>
- Clear list of blockers (must-fix to proceed).<br>
- Clear list of assumptions and validation needs for prototyping.<br>
- Any areas that likely require upstream contributions to Symfony AI.<br>
- Keep this work at the “rough plan” level: enough detail to guide the prototype phase, but not an implementation task.</p>
<p>- <strong>This issue was created with the help of AI</strong></p>
> Related issue: [Issue #3538479](https://www.drupal.org/node/3538479)
> Related issue: [Issue #3485449](https://www.drupal.org/node/3485449)
> Related issue: [Issue #3538499](https://www.drupal.org/node/3538499)
issue