Solution

Media & publishing

Drupal platforms for newsrooms and publishers: hundreds of stories a day, an archive that stays searchable, and pages that AI engines cite correctly.

A publisher's platform is an editorial workflow, an archive and a front door to search and AI engines, all at once. We build it on Drupal for high-volume editorial teams, keep the archive searchable and citable for a decade, and structure every story so the engines that now sit between you and your readers quote you right.

Talk about your publishing platform
  • Editorial workflow for hundreds of stories a day
  • Archives that stay searchable and citable
  • Structured data so AI answers quote you correctly

Standards and tools:Drupalschema.orgllms.txtWCAG 2.2AWS

What a publishing platform has to do

Move a story from a reporter to the front page in minutes, through an editor and a lawyer when it matters, and not lose it when the desk changes shift. Serve the front page in a fraction of a second to everyone at once when the story breaks. Keep ten years of archive findable, by a reader, by Google and now by an AI engine that will cite whoever it can parse. And do it under an accessibility standard, because a news site is a public accommodation.

What we do for publishers

We build and run Drupal platforms for newsrooms and publishers: a content model for stories, sections, authors, topics and editions; an editorial workflow with the review steps you actually need and none you do not; and an archive with the structure to stay searchable. We put schema.org data on every story, canonical URLs across every edition, an llms.txt file at the root, and the performance work that keeps Core Web Vitals green under load. We run it in your own cloud account, and we stay to patch and watch it.

How we work with newsrooms

An audit of the workflow as it is, the archive as it is, and the numbers: what search and AI engines cite today, how fast the front page is at peak. Then a content model designed with the desk, a migration rehearsed on the whole archive, and a cutover on a quiet day with every old URL redirected. After launch, monitoring that pages a person when the front page slows, and a monthly check on what gets cited.

Why now

AI answer engines are a new front door to your reporting, and they cite the pages they can parse, not the best ones. Interaction to Next Paint has been the responsiveness metric in Core Web Vitals since March 12, 2024, and it punishes heavy ad-laden pages. The Drupal AI module reached 1.4.8 in September 2026, which puts editorial AI, tagging, summaries and alt text at scale, inside the CMS. Drupal 10 ends on December 9, 2026.

The numbers behind a publishing platform

200 ms

INP for a "good" page

Interaction to Next Paint, a Core Web Vital since March 12, 2024. Heavy pages fail it.

INP on web.dev

Sep 3, 2024

llms.txt proposed

A file at the site root that tells AI agents what you publish. Every platform we build ships one.

The llms.txt proposal

1.4.8

Drupal AI module, current release

Released September 2, 2026: tagging, summaries and alt text at scale, inside the CMS.

Drupal AI on drupal.org

Dec 9, 2026

Drupal 10 end of life

Planned to coincide with the release of Drupal 12.

Drupal core release schedule

I want to

get cited by AI answer engines

When an AI engine covers your beat, it should cite your story and quote it accurately. That takes schema.org on every story, canonical URLs, plain structure and an llms.txt file.

  • NewsArticle and Article structured data on every story, checked on every build
  • Canonical URLs across editions, syndication and AMP leftovers
  • A monthly check on what gets cited, in writing

make the archive searchable again

Ten years of stories with inconsistent tags, missing summaries and dead images can be structured, described and indexed, with people checking the model's work at the rate they can manage.

  • A content model for stories, topics, authors and editions
  • Tagging and summaries drafted by a model, reviewed by the desk
  • Search that returns the right story, not the most recent one

hold the front page when the story breaks

The front page served in a fraction of a second to everyone at once, from the edge, with the CMS untouched by the crowd.

  • Front page and section pages cached at the edge, invalidated on publish
  • Load tested at the peak you have seen, then double
  • Core Web Vitals green under load, INP included

fix the editorial workflow

The review steps a newsroom actually needs, legal and standards when it matters, nothing in the way when it does not, with a story never lost between shifts.

  • Workflow designed with the desk, not for it
  • Roles for reporters, editors, legal and standards
  • An audit trail for every published change

put editorial AI to work safely

Tags, summaries, alt text and translations drafted inside the CMS, in your own cloud account, and never published without a person saying so.

  • Provider in your own cloud account, under your keys
  • Drafts land as suggestions, never as published content
  • Acceptance rate measured, so the feature earns its keep

get off an old Drupal without losing a URL

A migration of the whole archive, rehearsed until the real run is the boring one, with every old URL redirected so no ranking is lost.

  • An audit of modules, integrations and the archive's shape
  • A rehearsed migration of every story and image
  • Redirects for every URL, checked before cutover
How it runs

A publishing platform, from workflow audit to a cited archive

Decisions above the line, the steps that run on every build below.

01 A decision, made with you: Audit: workflow, archive, what gets cited, speed at peak
02 A decision, made with you: Content model designed with the desk
03 Runs on every build: Platform built with structured data, canonical URLs, llms.txt
04 Runs on every build: Archive migrated and rehearsed, every URL redirected
05 A decision, made with you: Desk trained, workflow signed off, then live
06 Runs on every build: Edge caching, monitoring, monthly citation check

The cutover happens on a quiet day. The archive is checked story by story before it, not after.

What is changing

What is changing for publishers

Checked on llmstxt.org, web.dev and drupal.org on September 12, 2026.

  1. AI engines are a front door

    They cite the pages they can parse. Structured, canonical, plainly written stories get quoted; the rest get summarized without a link.

  2. Responsiveness is measured

    Interaction to Next Paint has been a Core Web Vital since March 12, 2024, and 200 milliseconds is the bar. Heavy, script-laden pages fail it.

  3. Editorial AI moved into the CMS

    The Drupal AI module, at 1.4.8 since September 2, 2026, drafts tags, summaries and alt text at scale, inside your own cloud account, for a person to approve.

  4. Drupal 10 ends in December

    December 9, 2026, alongside the planned release of Drupal 12. An archive migration takes longer than a quarter.

Services for publishers

The three most newsrooms buy first.

  • Drupal development

    New builds, major-version upgrades, module and theme work, and migrations from Drupal 7 onward. Every site we build is one we can keep patched, accessible and on a supported version for years, because that is the job after launch.

  • AI engineering

    We put language models to work inside Drupal on the jobs they do well: alt text, summaries, tagging, and answers from your own pages with citations. They run in your tenancy, your content trains no one else's model, and nothing publishes without a person saying so.

  • Cloud engineering

    We design and run the cloud your platform lives on, in your own account, as code, with monitoring, patching and a monthly cost review in writing. We do not resell hosting, and everything we build is documented so the next engineer can run it.

Tell us about the newsroom.

Stories a day, the size of the archive, what it runs on, and what is hurting. We come back with what we would do first and what it would take.

Talk about your publishing platform

Fields marked are required.

When would you like work on your project to begin? (required)
What type of project would you like to discuss? (required) Tick at least one

Would you like to join our fabulously insightful newsletter that will help you grow your business? (an answer is required)

By joining you agree to receive marketing communications from Dynamosys. You can unsubscribe anytime. We won't share or sell your personal information. Privacy policy

Never share sensitive information (credit card numbers, social security numbers, passwords) through this form. This site is protected by Cloudflare's bot check.