A review-gated AI translation layer for multilingual Webflow
Translated pages that don't quietly go stale
Duplicating Webflow content across languages creates drift by design: the English source page changes, the translated copy sits untouched, and canonical or hreflang rules become nearly impossible to audit once enough pages exist.
Build translation as a reviewed pipeline, not a copy-paste task
Consolidate a multilingual proxy architecture that extracts English source HTML into stable, hashable translation segments, tracks their state in D1 through draft, reviewed, published, stale, and rejected, and prepares localized route output behind a Worker.
An AI translation proxy with a review gate before anything publishes
The architecture gives SoFlow a real multilingual publishing system - but it's staying private until storage, secrets, route ownership, and language policy are locked down.
Our other projects
The brief, self-directed, was to avoid maintaining duplicated translated Webflow pages by hand while still preparing SoFlow for multilingual publishing across language-prefixed routes.
Duplicating Webflow content across languages creates drift: source pages change, translated pages go stale, and canonical or hreflang rules become difficult to audit at scale.
This is not deploy-ready. D1 and KV bindings, provider secrets, the supported-language list, review policy, and production route ownership all still need final decisions before any public use.
SoFlow consolidated a multilingual proxy architecture that extracts English source HTML into stable translation segments, stores translation state in D1, supports draft, reviewed, published, stale, and rejected states, and prepares localized route output behind a Worker.
English Webflow pages remain the single source of truth. A Cloudflare Worker extracts route content, hashes each segment, flags stale rows for retranslation, batches translation through an AI provider adapter, and stores review state in D1 - with canonical, hreflang, and localized-route serving to follow once the renderer is wired in.
- English-source Webflow content extraction
- Source-hash based stale-translation detection
- D1-backed translation review states
- Protected review and translation diagnostics endpoints
- Canonical, hreflang, sitemap, and robots groundwork
The result is a serious multilingual publishing architecture for SoFlow's own site - deliberately kept private until storage, secrets, route ownership, and language policy are finished.
- Translation review module supports draft, reviewed, published, stale, and rejected states
- Protected endpoints exist for translation status, stale rows, page review, and review updates
- Project README explicitly marks the Worker as not deploy-ready
Is this proxy meant to be reused across future SoFlow language launches?
That's the intent. The architecture is deliberately generic - English source extraction, hashed segments, and a review-state machine - so adding a new supported language should mean a policy decision, not rebuilding the proxy from scratch.
Why put this in Webflow if external code is involved?
Webflow is the public storytelling and CMS layer. External code should stay in the app, Worker, or integration layer where it can be versioned, secured, and tested.
What has to happen before this goes live?
D1 and KV bindings, provider secrets, the supported-language list, route ownership, and renderer integration all need to be finalized before any language-prefixed route is exposed publicly.
This proves SoFlow treats Webflow localization as a controlled publishing system with review gates and staleness tracking, not just a page-copying exercise.
Is this proxy meant to be reused across future SoFlow language launches?
That's the intent. The architecture is deliberately generic - English source extraction, hashed segments, and a review-state machine - so adding a new supported language should mean a policy decision, not rebuilding the proxy from scratch.
Why put this in Webflow if external code is involved?
Heading 1
Heading 2
Heading 3
Heading 4
Heading 5
Heading 6
Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
Block quote
Ordered list
- Item 1
- Item 2
- Item 3
Unordered list
- Item A
- Item B
- Item C
Bold text
Emphasis
Superscript
Subscript
Heading 1
Heading 2
Heading 3
Heading 4
Heading 5
Heading 6
Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
Block quote
Ordered list
- Item 1
- Item 2
- Item 3
Unordered list
- Item A
- Item B
- Item C
Bold text
Emphasis
Superscript
Subscript
Multilingual publishing would have meant duplicated Webflow content or manual translation workflows prone to drifting from the English source.
There is now a clear Worker architecture for AI-assisted multilingual publishing, with review gates standing between a translation and anything going public.
A static Webflow localization setup can publish translated content, but on its own it can't track source-content hashes, flag translations as stale, keep unreviewed drafts from rendering, or expose review state per content segment.
