[arpanjot-coding]

The existing product site combines inconsistent app screenshots with claims the implementation does not support, including schema-only AI requests, zero hallucinations, a fabricated aggregate rating, and a fully rendered dashboard. This redesign gives Datly a consistent product presentation and describes the capabilities and limits found in the source.

The site now includes twelve static language pages: English, Russian, Simplified Chinese, Japanese, Vietnamese, Thai, Hindi, Arabic, French, German, Spanish, and Italian. Each has translated body copy and metadata, a self-canonical URL, reciprocal hreflang links (including x-default), a sitemap entry, and an explicit language selector. Arabic uses RTL layout; interactive gallery and filter messages use the same locale dictionaries.

• Adds five product compositions from repository application captures, downloadable SVG/JPG/WebP/2× PNG assets, a social card, and an image library with provenance. The report composition is explicitly labelled as an archived Data Formulator example.
• Places verified project statistics directly in the hero before app imagery: 249 GitHub stars, 16 data source options, 12 website languages, and MIT licensing. GitHub stars refresh from the public API with a dated fallback; figures link to their supporting details.
• Replaces unsupported adoption/security/performance/comparison claims with source-backed copy. Cloud model requests can contain sample values and rows; reports embed chart images; the current dashboard renders chart-reference placeholders.
• Updates the README and setup instructions, including the correct default port, 5567. The website stays buildless; locale and image authoring scripts are included.

Validation: all 12 pages pass translation-key coverage (220 entries each), placeholder preservation, language/direction, canonical/hreflang, sitemap, IDs, local link/asset, JSON, and download checks. JavaScript syntax and Git whitespace checks pass. Browser checks cover desktop/mobile rendering, French interactive translations, Arabic RTL and image dialogs, language selection, source filters, and setup controls.

Limits: translations have not had independent native-speaker editorial review. App backend, live external connectors, and desktop binaries were not end-to-end tested; this PR changes website/documentation and media, not application behavior. No production merge is performed.