Establish a localization baseline, monitor quality over time, and ship fixes faster — with AI-powered scanning, multi-agent capture, and a visual review workbench built for linguistic QA teams.
Paste a URL · Discover locales · Capture screenshots · Get full report
HOW GTW2 WORKS
From URL to full localization QA — in minutes
1 · URL & Discover
Target URL
https://experienceleague.adobe.com/
Reachable
🔍 Discover locales
Lingo agent: Discovering…
Scanning hreflang, sitemap & URL patterns… ⟳ ⟳ ⟳
2 · Review & Configure
Discovery results
Found 7 locale(s) · 15 page(s) · 7.0s
🇺🇸 EN base🇫🇷 FR🇩🇪 DE🇪🇸 ES🇮🇹 IT🇳🇱 NL🇯🇵 JA
Pages to scan
overviewSelected
home tutorials
introduction
7 locale(s) · 1 page(s) · ~8 screenshots
3 · AI Agents Scan
Agent squad · 45%
Capturing · IT
✓ Crawl · Detect · Setup
⟳ Capture · DE · ES · IT…
… Analyze · Done
Pages found1
Screenshots5/8
Issues—
4 · Scan Report
Scan complete ✓
1
Pages
8
Screenshots
19
Issues
Worst locale🇮🇹 IT (8)
Most commonUntranslated
Open in Workbench →
Export PDF
Scan and get your QA baseline
Run a scan to capture screenshots across all locales and surface localization issues — with locale-level severity breakdown.
Review with AI-assisted workbench
Annotate screenshots, classify bugs, and push to Jira — all in one place. AI pre-classifies severity so reviewers focus only on what matters.
Know your full release readiness
Live KQS score per language, bug trends, team velocity, and client health — all in the dashboard before every release.
Step 1 · URL & Discover
Start with a live localization baseline in one scan
Enter any website URL and GTW2 crawls key pages with full JavaScript rendering to establish your localization baseline. Within minutes you get locale detection, page discovery, and screenshot estimates — ready to scan.
PRODUCT PREVIEW
…/localization-scan/new
1
URL
2
Review
3
Scan
G T W 2
Localization QA scanner
Paste a URL and we'll discover available locales, crawl pages, and let you pick exactly what to scan.
Scout
Crawler
Lingo
Locale detector
Pixel
Screenshot agent
Hawk
Visual analyzer
Ops
Platform agent
Atlas
Report builder
TARGET URL
Reachable
Discovering locales…
Scanning hreflang, sitemap & URL patterns
Set URL & discover locales
AI locale detection
Start with a live localization baseline in one scan
Enter any website URL and GTW2 crawls key pages with full JavaScript rendering. The Lingo agent scans hreflang tags, sitemaps, and URL patterns to auto-detect every locale — no manual config needed.
Auto-detects path-prefix, subdomain, and query-based locale strategies
Finds 7+ locales and 15 pages in under 7 seconds
Handles heavy SPAs — full JS rendering
Track issue volume and locale health from the very first scan
Track issue volume, page impact, and locale health from the very first scan.
Localization QA scanner
TARGET URL
https://experienceleague.adobe.com/● Reachable
Scout
Lingo
Pixel
Hawk
Ops
Atlas
Discovering locales…
Scanning hreflang, sitemap & URL patterns
Open full-screen GTW2 UI prototype (simulated data)
Zero configuration needed
No need to manually list your locales. GTW detects path-prefix, subdomain, and query-based strategies automatically — just paste the URL.
Discover pages in seconds
Finds 7 locales and 15 pages in under 7 seconds. You see every page available to scan and can select exactly which ones to include before starting.
Full JS rendering
Heavy SPAs products can take 1–3 min to render. GTW waits for full render before capturing — ensuring screenshots reflect the real page state.
Step 2 · Review & Configure
Pick your pages and locales — then confirm in one click
After discovery, you see every locale and page found. Select exactly what to scan — one page or fifteen — and choose whether to create a new product or link to an existing one. GTW handles the rest.
PRODUCT PREVIEW
…/localization-scan/new — Review
Discovery results
Found 7 locale(s) and 15 page(s) in 7.0s
URL strategy:path-prefixBase:EN
Available locales
🔒 🇺🇸 EN base🇫🇷 FR MEDIUM🇩🇪 DE MEDIUM🇪🇸 ES MEDIUM🇮🇹 IT MEDIUM🇳🇱 NL 🇯🇵 JA 🇰🇷 KO
Pages to scan
Select all · Clear
overview
experienceleague.adobe.com/en/docs/…/overview
home tutorials
introduction
aa overview
7 locale(s) · 1 page(s) · ~8 screenshots
Continue →
Select locales & pages to scan
Estimated screenshots calculated
Smart page & locale selection
Control exactly what gets scanned — no waste
GTW shows you every discovered page and locale. Select one page or all fifteen. Pick EN + DE + IT only or all 8 locales. Preview the estimated screenshot count before you commit. Full control, zero surprises.
URL strategy auto-displayed — path-prefix, subdomain, or query
Pages you selected during discovery are pre-loaded into the scan. No re-entering URLs — the workflow remembers your choices from step to step.
Link to existing product
Already have a client for this hostname? Link this scan to an existing job — all historical data, bugs, and screenshots stay connected under one product.
Screenshot preview count
Before you start, GTW shows exactly how many screenshots will be taken. 7 locales × 1 page = 8 screenshots. Predictable usage, no surprise quotas.
Step 3 · AI Scan in Progress
Six AI agents work in parallel — scan to report
Once started, GTW's agent squad runs simultaneously across all stages. Scout crawls, Lingo resolves URLs, Pixel captures screenshots, Hawk analyzes, Ops saves assets, and Atlas compiles the report — all without you lifting a finger.
PRODUCT PREVIEW
…/localization-scan/[id]
G T W 2 — https://experienceleague.adobe.com/
Scan complete
✓ Crawl✓ Detect✓ Setup✓ Capture✓ Analyze✓ Done
Complete100%
experienceleague.adobe.com/
IT ✓ Saved
Activity
▶ 12:01:53 ES: Loaded start URL
▶ 12:01:49 NL: Taking screenshot…
▶ 12:01:34 DE: Taking screenshot…
▶ 12:01:24 DE: Waiting for render…
▶ 12:01:00 Scan complete
1
Pages discovered
8
Screenshots
19
Issues found
Agent squad — all complete
Scout
Done ✓
Lingo
Done ✓
Pixel
Done ✓
Hawk
Analysis complete ✓
Ops
Workspace updated ✓
Atlas
Report ready ✓
View report →Scan another URLRescan
Six-agent parallel execution
From crawl to report — fully automated
GTW's six agents run simultaneously across pipeline stages. While Pixel is still capturing IT screenshots, Ops is already saving EN assets and Hawk is queuing analysis. True parallelism means no waiting between steps.
▶ 15:34:10 Capturing overview / DE… ▶ 15:34:10 Capturing overview / ES… ▶ 15:34:11 JA: Waiting for page to render…
Real-time activity log
Every agent action is logged in real time — timestamp, locale, action. You can watch the scan progress and know exactly which locale is being captured at any moment.
Parallel browser execution
8 browser instances capture screenshots simultaneously via Apify Cloud. Parallel limit queuing ensures no locale is skipped — even under high concurrency.
Stop & rescan anytime
Stop a scan mid-way if needed. All captured screenshots up to that point are preserved. Rescan from the same page to pick up where you left off — no data lost.
Step 4 · Scan Report
Full scan report — visual comparison, issues, export ready
The scan report gives you everything: per-locale screenshot comparison, issue count by severity, worst locale identification, and export to PDF.
PRODUCT PREVIEW
…/localization-scan/[id]/report
GTW2 · Localization / Scan / Report
Scan report
Export PDF
Open in Workbench
Re-run AI analysis
Pages
1
Locales
8
Screenshots
8
Issues
19
Critical
⚠ 19
Platform setup
Clientexperienceleague.adobe.com
JobLocalization scan 2026-04-15 06:30
Scan date15/04/2026, 12:00:45
Duration167s
Results & insights
Issues by severity: Critical 19Warning 0
Most common type:Untranslated
Worst locale: 🌐 IT (8 issues)
✓ Screenshots saved as assets on the project.
Visual comparison Issues 19 Export
Page
🇺🇸 EN
✓ clean
🇫🇷 FR
✓ clean
🇩🇪 DE
5 issues
🇪🇸 ES
6 issues
🇮🇹 IT
8 issues
🇳🇱 NL
✓ clean
📄 /docs/certification/program/overview · 19 total · all locales
Scan report & visual comparison
See every locale side by side — find the worst instantly
The scan report grid shows EN source alongside every target locale. Clean locales get a green badge. Locales with issues show a red count. Click any locale to open it in the workbench and start reviewing immediately.
Visual comparison grid — EN vs FR, DE, ES, IT, NL, JA, KO
Worst locale auto-identified — IT (8 issues) flagged immediately
Most common issue type — Untranslated strings surfaced upfront
Export full PDF report for client delivery and stakeholder review
See EN source alongside every target locale in the same row. Green "clean" badges vs red issue counts make it instantly clear which locales need attention.
Worst locale auto-flagged
GTW auto-identifies the worst locale (IT with 8 issues) and most common bug type (Untranslated) so your team knows exactly where to start reviewing first.
One-click workbench open
Click any locale thumbnail to open it directly in the review workbench with source and target side-by-side, annotation tools ready, and the bug composer pre-loaded.
Step 5 · Review Workbench
Review every locale — annotate, classify, and ship to Jira
The workbench puts source and target screenshots side by side. Draw annotation boxes directly on the screenshot, describe the issue in plain language, set severity, and submit.
Turn violations into sprint-ready engineering tasks
GTW turns raw localization bugs into prioritized workstreams. The split-view workbench puts EN source alongside your target locale — annotate, describe in plain language, and submit. Move from scattered issues to a clear P1 → P2 remediation roadmap.
Side-by-side EN source vs any target locale
Draw annotation boxes directly on screenshots to flag issues
AI pre-classifies severity — reviewers confirm, not classify
One-click Jira sync with full context — screenshot, severity, locale
Move from scattered issues to a clear P1 → P2 remediation roadmap.
Upload source and target files (JSON, XLIFF, PO, CSV, TXT) and start reviewing immediately. No client or product setup required — zero friction for one-off reviews.
AI severity pre-classification
GTW's AI automatically classifies each bug as Critical, Major, Minor, or Cosmetic before your reviewer opens the screen — so reviewers focus on confirming, not classifying.
Deadline & completion tracking
Every locale has a deadline and completion status. "Mark French Complete" closes the review loop. 14 days left shown at all times — no missed SLAs.
Step 6 · Live Dashboard
Real-time visibility into your entire localization operation
The GTW dashboard gives every stakeholder — QA leads, PMs, client managers — a single view of coverage, quality scores, bug trends, and release readiness across all languages and products.
PRODUCT PREVIEW
…/dashboard
Dashboard
Overview of your workspace metrics
7d30d90d
22
Total Clients
↑ 100%
23
Active Products
↑ 100%
27
Open Projects
↑ 1100%
5
Screens Reviewed
4
Bugs Found
↑ 200%
52
AI Flags
↓ 76%
Quality Trend
7d30d
Bug Distribution
4
Total
● Critical 1● Major 3● Minor 0
Language Quality Matrix
Lang
Screens
AI Flags
Bugs
Review %
🇪🇸 ES
5
29
0
0%
🇰🇷 KO
2
0
0
0%
🇮🇳 HI
15
10
2
67%
🇬🇧 EN
16
2
1
6.67%
Monitor health over time
Live QA dashboard
No more flying blind — know your release readiness
GTW's live dashboard gives you a single number for every language's localization health. Track trends, catch regressions, and make confident go/no-go decisions backed by data — not gut feel. Every stakeholder sees the same truth.
Language quality matrix — screens, AI flags, bugs, review % per locale
Quality trend chart — spot regressions over 7d, 30d, 90d windows
Bug distribution donut — Critical, Major, Minor at a glance
Track your localization quality score across 7d, 30d, and 90d windows. Catch regressions the moment they appear — before your clients do.
Client health scores
Every client has a health score based on their bug trends, coverage rates, and review velocity. Proactively manage relationships before escalations happen.
Release readiness indicator
Before every release, the dashboard shows green / amber / red readiness per language. Make confident go/no-go decisions backed by real data — not gut feel.
Start your localization QA audit today
Scan your website free. Get locale detection, full screenshot comparison, AI-flagged issues, and a sprint-ready fix plan — in under 3 minutes. No credit card required.
