Adds explanation on when to use dense or sparse embeddings

[kosabogi]

Summary

This PR adds explanation on when to use dense and sparse vectors to the Tutorial: Dense and sparse workflows using ingest pipelines page.

Related issue: https://github.com/elastic/docs-content-internal/issues/854

Generative AI disclosure

1. Did you use a generative AI (GenAI) tool to assist in creating this contribution?

• Yes
• No

Claude in Cursor

[github-actions]

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]

🔍 Preview links for changed docs

• solutions/search/vector/dense-versus-sparse-ingest-pipelines.md

[seanhandley]

Looks almost ready @kosabogi - just a couple of thoughts.

[github-actions]

Elastic Docs Style Checker (Vale)

Summary: 1 suggestion found

💡 Suggestions (1): Optional style improvements. Apply when helpful.

File	Line	Rule	Message
solutions/search/vector/dense-versus-sparse-ingest-pipelines.md	55	Elastic.FirstPerson	Use caution when using first-person pronouns such as 'my.'

---

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.

[kosabogi]

Thanks for the feedback, @seanhandley and @leemthompo! I’ve added context engineering as a use case for dense vectors. Let me know how it looks now.

[leemthompo]

Nice! I wonder if this is the right spot to mention that you can also combine these approaches. A production search setup might combine lexical, dense, and sparse in a tiered approach. But this page is set up as a "versus" so perhaps food for a follow-up...