Skip to content

Clarify Vector Database project guidance - #8092

Open
kosabogi wants to merge 6 commits into
mainfrom
vector-database-project-context
Open

Clarify Vector Database project guidance#8092
kosabogi wants to merge 6 commits into
mainfrom
vector-database-project-context

Conversation

@kosabogi

@kosabogi kosabogi commented Aug 26, 2026

Copy link
Copy Markdown
Member

Summary

  • Add the Vector Database project type and exact value to cross-project search tag guidance.
  • Clarify the Elasticsearch and Vector Database project choices in the Search getting-started flow.
  • Distinguish Elasticsearch Serverless API profiles from the dedicated Vector Database project type using tabs.
  • Recommend the Vector Database project type for embedding-driven workloads on the vector search landing page.

This draft follows the project positioning introduced in #7961. Open product questions are listed in a separate PR comment.

Distinguish the dedicated project type from Elasticsearch profiles and surface it where users select projects or route cross-project searches.

Co-authored-by: Cursor <cursoragent@cursor.com>
@kosabogi

kosabogi commented Aug 26, 2026

Copy link
Copy Markdown
Member Author

Questions to confirm

@alexmarhaba @dustincoates Before finalizing the content, could you please confirm the following?

  1. Project selection: Is this the intended recommendation: use Vector Database when embeddings and similarity search are central to the workload, and use the Elasticsearch project type for general-purpose or mixed lexical, time series, and analytics workloads?
  2. Agent Builder: I can see Agent Builder in a Vector Database project in QA, and the current Kibana configuration also enables it and includes it in the Vector Database navigation. However, PR #7961 currently tells users to choose the Elasticsearch project type when they need Agent Builder. Is that guidance outdated? Should Agent Builder be removed from that Elasticsearch-specific recommendation, and should we document a Vector Database-specific access path?

@github-actions

Copy link
Copy Markdown
Contributor

Elastic Docs AI PR menu

Check the box to run an AI review for this pull request.

  • Review docs changes (docs-review). Status: not started.

Powered by GitHub Agentic Workflows and docs-actions. For more information, reach out to the docs team.

@github-actions

github-actions Bot commented Aug 26, 2026

Copy link
Copy Markdown
Contributor

@github-actions

Copy link
Copy Markdown
Contributor

✅ Elastic Docs Style Checker (Vale)

No issues found on modified lines!


The Vale linter checks documentation changes against the Elastic Docs style guide. To use Vale locally or report issues, refer to Elastic style guide for Vale.

Use the API values readers need when filtering or routing by the predefined project type tag.

Co-authored-by: Cursor <cursoragent@cursor.com>
Separate Elasticsearch profiles from the Vector Database index mode so readers can select the correct configuration for their project type.

Co-authored-by: Cursor <cursoragent@cursor.com>
kosabogi and others added 2 commits August 26, 2026 11:41
Surface the separate project type and its automatic vector index mode where readers first learn about vector search.

Co-authored-by: Cursor <cursoragent@cursor.com>
Recommend the Vector Database project for embedding-driven Serverless workloads without duplicating lower-level configuration details.

Co-authored-by: Cursor <cursoragent@cursor.com>
@alexmarhaba

Copy link
Copy Markdown

Should Agent Builder be removed from that Elasticsearch-specific recommendation, and should we document a Vector Database-specific access path?

Agent builder is available on all project types, so not sure why we're recommending to use ES for it here, I would remove that mention.

@kosabogi
kosabogi marked this pull request as ready for review August 31, 2026 13:18
@kosabogi
kosabogi requested a review from a team as a code owner August 31, 2026 13:18
@kosabogi

Copy link
Copy Markdown
Member Author

2. PR #7961 currently tells users to choose the Elasticsearch project type when they need Agent Builder

Thanks Alex, I left a comment on that PR mentioning this.
cc @yetanothertw

@dustincoates

Copy link
Copy Markdown
Contributor

Project selection: Is this the intended recommendation: use Vector Database when embeddings and similarity search are central to the workload, and use the Elasticsearch project type for general-purpose or mixed lexical, time series, and analytics workloads?

That would be a good way to put it. We'll probably want to eventually change it so that more and more of the mixed workloads move over to VDB, but for now, I think this is good.

# Vector search in {{es}}

:::{tip}
If you're using {{serverless-full}}, the [{{es}} Vector Database project type](/solutions/vector-database.md) is recommended when embeddings and similarity search are central to your application.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We should avoid admonitions at the very top of a page as much as possible. This has been abused in the past. This info can go in a dedicated subsection, as high as you like in the tree. This TIP now has 3 different links, which is a recipe for immediate user confusion.

@leemthompo leemthompo left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We're now distributing decision guidance across get-started.md, vector.md, dense-vector.md.

Think about presenting this together on one neutral comparison page or in one expanded "Choose your deployment type" section in get-started.md. The vector overview and dense-vector pages could then link to that guidance instead of duplicating partial recommendations.

The page needs to compare:

  • Elasticsearch with general_purpose
  • Elasticsearch with vector
  • Elasticsearch VDB

A comparison table might be clearer than tabs because users need to evaluate all three choices together.

When you use the [API]({{cloud-serverless-apis}}operation/operation-createelasticsearchproject) to create an {{es-serverless}} project, you can choose between two profiles: `general_purpose` and `vector`.
The general purpose profile is recommended for most search use cases, including full text search, sparse vectors, and dense vectors that use compression such as BBQ.
The vector profile is recommended only when your use case is based on uncompressed dense vectors with high dimensionality.
The available configuration depends on your project type.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this information appears too late: users must choose a project before they can index and query data, but the section comes after those implementation steps.

and this isn't only about vectors so it also feels a bit misplaced

:::

If you're not sure which project type to choose, select the **{{es}}** project type. This project type provides core {{es}} search capabilities along with additional UI tools to help you build search-powered applications faster.
Select the project type that matches your workload:

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'd think the "Choose a {{serverless-short}} project configuration" should live alongside this information as they go hand in hand?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants