Add Plasmate integration#467
Open
dbhurley wants to merge 1 commit into
Open
Conversation
Plasmate is an open-source (Apache 2.0) browser engine for AI agents that produces the Semantic Object Model (SOM) — a flat, typed JSON document representing a web page in a form optimized for LLM consumption. This integration adds PlasmateWebFetcher and PlasmateSOMConverter components for Haystack 2.0 RAG pipelines as a drop-in alternative to LinkContentFetcher and HTMLToDocument, with ~17x average token reduction across the public WebTaskBench benchmark. Reopening the redirect from deepset-ai/haystack#11056 (closed Apr 13 with @anakin87 suggesting this venue). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
kacperlukawski
requested changes
May 15, 2026
Member
kacperlukawski
left a comment
kacperlukawski
left a comment
There was a problem hiding this comment.
Thank you, I left some comments.
In general, it seems the package does not exist. How did you test the code snippets? Installing the package should fail, as pypi does not recognize the name.
|
|
||
| [Plasmate](https://plasmate.app) is an open-source (Apache 2.0) browser engine designed from the ground up for AI agents. Instead of rendering pixels, Plasmate produces the Semantic Object Model (SOM) — a flat, typed JSON document representing a web page in a form optimized for LLM consumption. | ||
|
|
||
| Across 38 measured production sites, Plasmate achieves an average **~17× token reduction** versus raw HTML, with peaks above **100×** on large SaaS marketing pages. The reproducible benchmark is published at [webtaskbench.com](https://webtaskbench.com), and the SOM/1.0 format is specified at [somspec.org](https://somspec.org/spec). |
Member
There was a problem hiding this comment.
I would prefer to avoid putting the benchmark results in the integration page. They may simply get outdated soon, but we also don't want to give the reader an impression that we suggest using a specific tool. Moreover, the benchmark itself seems to be affiliated with Plasmate.
Comment on lines
+138
to
+140
| ## SOM directives — making your own pages Plasmate-friendly | ||
|
|
||
| If you publish content that you would like AI agents to read efficiently, advertise a SOM endpoint via [SOM Directives in robots.txt](https://somspec.org/directives). The five-line opt-in tells any compatible agent to fetch a structured representation of your pages instead of the full HTML rendering. Verify your site is SOM-ready at [somready.com](https://somready.com). |
Member
There was a problem hiding this comment.
It doesn't seem to be related to the integration itself. I suggest removing that part.
| - name: Plasmate Labs | ||
| socials: | ||
| github: plasmate-labs | ||
| pypi: https://pypi.org/project/haystack-plasmate |
Member
There was a problem hiding this comment.
The package does not seem to exist.
| repo: https://github.com/plasmate-labs/haystack-plasmate | ||
| type: Data Ingestion | ||
| report_issue: https://github.com/plasmate-labs/haystack-plasmate/issues | ||
| logo: /logos/plasmate.png |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds the Plasmate integration to the showcase.
Following @anakin87's suggestion in deepset-ai/haystack#11056 — that PR proposed a
PlasmateFetcherdirectly inside the Haystack core, which @anakin87 redirected here. The integration package itself already lives atplasmate-labs/haystack-plasmate(Haystack 2.0 components:PlasmateWebFetcherandPlasmateSOMConverter).What Plasmate is
Open-source (Apache 2.0) browser engine that produces the Semantic Object Model (SOM) — a flat, typed JSON representation of a web page optimized for LLM consumption. Drop-in alternative to
LinkContentFetcher/HTMLToDocumentthat delivers an order of magnitude lower token cost per page.Public benchmark (38 sites, weekly updated): https://webtaskbench.com — currently 29.6× average compression, 9.8× median, 118.5× peak (cloud.google.com).
Files changed
integrations/plasmate.md— integration markdown matching the existing pattern (apify.md, anthropic.md, etc.) with Overview / Installation / Components / RAG pipeline example / License sections.Logo
/logos/plasmate.pngis referenced but not added in this PR. Happy to add it in a follow-up commit if you let me know your size/format preference (most existing logos look like ~512px PNG with transparency). Or feel free to drop one in directly.Thanks for the steer to the right venue.