# Developer Fix Pack

You are my AI coding agent. Improve the homepage at https://www.agentui.ai/en/ using the audit below.

Goal: edit the homepage source code so the page is clearer, more specific, more trustworthy, and easier to convert. Do not just summarize the audit.

Before editing:
- Inspect the homepage implementation and reuse the existing framework, components, styles, and copy patterns.
- Keep the same product and offer. Change the framing, copy, layout, CTA support text, and metadata where needed.
- Do not invent pricing, plan limits, free-trial details, credit-card policy, cancellation terms, setup time, security claims, customer names, metrics, or guarantees.
- Use only facts present in the audit or the current page. If a fact is missing, use placeholders like [actual starting price] or [actual credit-card policy] and leave it obvious for me to fill in.
- Prefer concrete copy and UI changes over generic advice.

Make these changes:
1. Rewrite the hero section: headline, subheadline, primary CTA, secondary CTA if useful, and the short proof or reassurance line under the CTA.
2. Rewrite weak page sections called out in the audit with exact replacement copy.
3. Add or move trust proof, objection handling, and CTA support near the conversion points.
4. Apply visual hierarchy fixes from the audit: spacing, CTA contrast, section order, typography emphasis, and mobile density.
5. Fix SEO, load-speed, mobile, color/contrast, and technical issues where the code allows.
6. Keep the page honest: replace unsupported facts with placeholders instead of made-up examples.
7. Run the relevant formatter, linter, or tests after editing.

## Score Summary

- Headline & Subheadline: 4/10
- Value Proposition: 5/10
- CTA Architecture: 3/10
- Trust & Social Proof: 5/10
- Anxiety & Objection Handling: 4/10
- Copy Specificity & Tone: 6/10
- Page Narrative & Flow: 5/10
- Visual Hierarchy: 4/10
- Color & Accessibility: 3/10
- Friction & Distraction: 4/10
- Mobile Experience: 3/10
- SEO Fundamentals: 3/10
- Load Speed: 3/10
- Technical Health: 4/10


## Critical Fixes (do these first)

- **cro**: Primary CTA has no inspectable destination -> Use a real link or form submit path so visitors, crawlers, analytics, and QA can verify the conversion path
- **visual**: Sticky overlay takes too much mobile space -> Reduce or dismiss sticky bars or popups so the CTA and value prop remain readable
- **performance**: Render-blocking scripts detected -> Add defer or async where safe and move noncritical scripts below first render
- **accessibility**: Form controls lack labels -> Associate every input, select, and textarea with a visible label or aria-label


## Fix First Sequence

1. Add defer or async where safe and move noncritical scripts below first render. Validate: Re-run the audit and confirm this finding is resolved: Render-blocking scripts detected.
2. Set width and height or aspect-ratio to reduce layout shift. Validate: Re-run the audit and confirm this finding is resolved: Images are missing reserved dimensions.
3. Remove noncritical third-party tags and combine or defer low-value assets. Validate: Re-run the audit and confirm this finding is resolved: Too many network requests.
4. Remove low-value scripts and split noncritical code away from the landing page. Validate: Re-run the audit and confirm this finding is resolved: JavaScript file count is high.
5. Consolidate analytics, widgets, fonts, and embeds; keep only origins that directly support conversion. Validate: Re-run the audit and confirm this finding is resolved: Too many third-party origins.
6. Keep only conversion-critical third-party scripts on the landing page. Validate: Re-run the audit and confirm this finding is resolved: Too many third-party requests.


## Findings To Cover

- `performance.blocking_scripts` [critical / performance / confidence 0.9] Location: Lighthouse diagnostics. Evidence: 4 blocking scripts were extracted. Exact fix: Add defer or async where safe and move noncritical scripts below first render.
- `performance.image_dimensions` [high / performance / confidence 0.9] Location: Lighthouse CLS diagnostics. Evidence: 1 of 2 images lack width and height. Exact fix: Set width and height or aspect-ratio to reduce layout shift.
- `performance.request_count` [high / performance / confidence 0.9] Location: Semrush Performance report pattern. Evidence: 168 requests were captured. Exact fix: Remove noncritical third-party tags and combine or defer low-value assets.
- `performance.script_count` [high / performance / confidence 0.9] Location: Semrush JS/CSS report pattern. Evidence: 24 external scripts were extracted. Exact fix: Remove low-value scripts and split noncritical code away from the landing page.
- `performance.third_party_domain_count` [high / performance / confidence 0.9] Location: web.dev third-party JavaScript guidance. Evidence: 10 third-party domains were captured. Exact fix: Consolidate analytics, widgets, fonts, and embeds; keep only origins that directly support conversion.
- `performance.third_party_requests` [high / performance / confidence 0.9] Location: Lighthouse third-party impact pattern. Evidence: 56 third-party requests were captured. Exact fix: Keep only conversion-critical third-party scripts on the landing page.
- `accessibility.form_labels` [high / accessibility / confidence 0.86] Location: WCAG forms. Evidence: 1 of 1 controls lack a label or accessible name. Exact fix: Associate every input, select, and textarea with a visible label or aria-label.
- `visual.sticky_overlay_blocks_mobile` [high / visual / confidence 0.82] Location: mobile viewport. Evidence: Sticky overlay covers 72% of the mobile viewport. Exact fix: Reduce or dismiss sticky bars/popups so the CTA and value prop remain readable.
- `cro.primary_cta_no_destination` [high / cro / confidence 0.78] Location: Post-click funnel QA. Evidence: Primary CTA is "Book a Call" with no href. Exact fix: Use a real link or form submit path so visitors, crawlers, analytics, and QA can verify the conversion path.
- `visual.obvious_overlap` [high / visual / confidence 0.78] Location: rendered viewport. Evidence: 17 visible element overlaps were detected in the rendered viewport. Exact fix: Inspect stacked/absolute/fixed elements at mobile and desktop breakpoints and remove incoherent overlaps.

## Overall Assessment

Things done well:
- Strong differentiation between demo and working system in value proposition
- Sound familiar section uses excellent voice-of-customer language
- Numeric proof metrics exist such as 67 percent process efficiency
- Meta description promises human support which matches the product differentiator

Priority issues to fix:
1. Repair the broken primary CTA href and standardize all button copy to Start Building Free to eliminate conversion path fragmentation
2. Reduce the mobile sticky overlay from 72 percent to under 15 percent and increase tap targets to 44 pixels to restore mobile usability
3. Defer the 4 render-blocking scripts and audit the 56 third-party requests to improve load speed and Core Web Vitals

## Conversion Theory

Visitors bounce because the primary call-to-action button does not link anywhere, mobile users cannot see the value proposition due to a massive sticky overlay covering 72 percent of the screen, and the headline defines the product by what it is not rather than what it does.


## Tool-Style Checks

These objective checks are modeled after technical SEO, local browser lab, accessibility, and crawl-audit tools. Treat critical and high severity issues here as hard requirements.

- [critical] Render-blocking scripts detected: 4 blocking scripts were extracted. Fix: Add defer or async where safe and move noncritical scripts below first render.
- [high] Form controls lack labels: 1 of 1 controls lack a label or accessible name. Fix: Associate every input, select, and textarea with a visible label or aria-label.
- [high] Primary CTA has no inspectable destination: Primary CTA is "Book a Call" with no href. Fix: Use a real link or form submit path so visitors, crawlers, analytics, and QA can verify the conversion path.
- [high] Images are missing reserved dimensions: 1 of 2 images lack width and height. Fix: Set width and height or aspect-ratio to reduce layout shift.
- [high] Too many network requests: 168 requests were captured. Fix: Remove noncritical third-party tags and combine or defer low-value assets.
- [high] JavaScript file count is high: 24 external scripts were extracted. Fix: Remove low-value scripts and split noncritical code away from the landing page.
- [high] Too many third-party origins: 10 third-party domains were captured. Fix: Consolidate analytics, widgets, fonts, and embeds; keep only origins that directly support conversion.
- [high] Too many third-party requests: 56 third-party requests were captured. Fix: Keep only conversion-critical third-party scripts on the landing page.

## Ranked Action Plan

Use this impact/effort/confidence ranking before choosing what to edit first:

- Score 4.86: Images are missing reserved dimensions (optimize, effort 2/5, confidence 90%). Fix: Set width and height or aspect-ratio to reduce layout shift.
- Score 4.86: Too many network requests (optimize, effort 2/5, confidence 90%). Fix: Remove noncritical third-party tags and combine or defer low-value assets.
- Score 4.86: JavaScript file count is high (optimize, effort 2/5, confidence 90%). Fix: Remove low-value scripts and split noncritical code away from the landing page.
- Score 4.72: Render-blocking scripts detected (optimize, effort 4/5, confidence 90%). Fix: Add defer or async where safe and move noncritical scripts below first render.
- Score 4.68: Primary CTA has no inspectable destination (test, effort 2/5, confidence 78%). Fix: Use a real link or form submit path so visitors, crawlers, analytics, and QA can verify the conversion path.
- Score 4.28: Form controls lack labels (fix, effort 2/5, confidence 86%). Fix: Associate every input, select, and textarea with a visible label or aria-label.


## Opportunity Clusters

Treat these as root-cause workstreams. Avoid fixing one symptom while leaving the related blockers untouched:

- Score 8.85: Speed and Delivery (9 related findings, metric: mobile_bounce_rate). Lead fix: Set width and height or aspect-ratio to reduce layout shift.
- Score 5.93: Site Quality (6 related findings, metric: site_health_score). Lead fix: Use a real link or form submit path so visitors, crawlers, analytics, and QA can verify the conversion path.
- Score 4.99: Accessibility and Usability (3 related findings, metric: usable_conversion_rate). Lead fix: Associate every input, select, and textarea with a visible label or aria-label.
- Score 4.43: Indexability and Discoverability (3 related findings, metric: indexed_click_eligible_pages). Lead fix: Add a self-referencing canonical URL for this landing page.
- Score 2.35: Conversion Path (1 related findings, metric: primary_cta_click_rate). Lead fix: Reduce above-fold navigation to the few paths visitors need to decide.


## Recommendation Briefs

Use these as the implementation plan. Keep validation events and metrics intact:

- Speed and Delivery [optimize]: Set width and height or aspect-ratio to reduce layout shift. Validate with mobile_bounce_rate.
- Site Quality [fix_now]: Use a real link or form submit path so visitors, crawlers, analytics, and QA can verify the conversion path. Validate with site_health_score.
- Accessibility and Usability [fix_now]: Associate every input, select, and textarea with a visible label or aria-label. Validate with usable_conversion_rate.
- Indexability and Discoverability [test]: Add a self-referencing canonical URL for this landing page. Validate with indexed_click_eligible_pages.
- Conversion Path [test]: Reduce above-fold navigation to the few paths visitors need to decide. Validate with primary_cta_click_rate.


## Stack, Form, and Trust Signals

- Next.js: framework
- Forms: 0 | Fields: 0 | Required: 0 | Missing autocomplete: 0 | Business fields: 0 | Sensitive fields: 0 | Confirm fields: 0 | Privacy microcopy: no | Social auth: 0
- Privacy: True | Terms: True | Security: True | Pricing: True | Logos: 0 | Testimonials: 0 | Case studies: 0 | Risk reversal: True
- Analytics: ga4, meta_pixel | Events: purchase | UTM fields: 0 | Params: 1 | Form starts: 0 | CTA clicks: 0 | Form errors: 0 | GA IDs: 0 | Consent tools: termly
- Schema types: Organization, WebSite | Items: 2 | Invalid JSON-LD: 0
- Hreflang alternates: 12 | Self-reference: True | x-default: False | Invalid codes: 0
- URL length: 26 | Depth: 1 | Params: 0 | Canonical path mismatch: False
- Images: 2 | Responsive: 0 | Legacy: 1 | Videos: 0
- Font stylesheets: 0 | Font preloads: 1 | Missing font-display: 0 | Legacy fonts: 0
- Compression: br | Cache-Control: True | Validators: True | CDN cache signal: True
- Hero CTA: True | Hero words: 84 | CTA count: 20 | Footer CTAs: 16 | CTA destinations: 5 | External CTA paths: 5 | Missing CTA anchors: 0
- Section flow: 12 sections | Sequence: hero, how_it_works, problem, proof, solution, how_it_works, proof, problem, solution, objections, generic, generic
- Main landmarks: 1 | Generic links: 0 | Small targets: 0 | Unsafe new-tab links: 10


- Competitor matrix: keyword-aligned competitors 0 | competitors with high-intent CTA 0 | competitors with proof 0 | missing terms: none


## Social Profiles To Review

Use these public profiles to cross-check positioning, proof, audience, trust, and launch-channel suggestions:

- linkedin: https://www.linkedin.com/company/agentui
- instagram: https://www.instagram.com/agent.ui

- Organization sameAs URLs: 3


## Experiment Ideas

- Test indexability and discoverability: Because Missing canonical URL. Search and AI systems must be able to crawl, understand, and trust the page. we believe Add a self-referencing canonical URL for this landing page. will improve indexed_click_eligible_pages. Primary metric: indexed_click_eligible_pages.
- Test conversion path: Because Navigation may be overloaded. Visitors need an obvious next step before secondary improvements can pay off. we believe Reduce above-fold navigation to the few paths visitors need to decide. will improve primary_cta_click_rate. Primary metric: primary_cta_click_rate.
- Reduce above-fold navigation choices: Reducing competing navigation choices will increase attention on the primary conversion path. Primary metric: primary_cta_click_rate.


## Detailed Breakdown Per Section

### Headline & Subheadline
**Score:** 4/10
**Issue:** The H1 reads 'It's not a demo.' which forces the visitor to guess what it IS. The subheadline 'It's the complete system.' is vague and does not mention the core mechanism (AI) or the target audience (companies/operations). The meta title says 'AI + Human App Builder' but this is missing from the visible H1.
**Fix:** Change H1 to 'Build Internal Business Apps with AI and Human Support.' Change subheadline to 'Stop prototyping. Get a working system with your data, users, and permissions in minutes.'
### Value Proposition
**Score:** 5/10
**Issue:** The hero copy states 'We deliver it working' which contrasts well with 'Other AI tools show you what... could look like.' However, the unique selling proposition 'Backed by humans' found in the H2 'Built by AI. Backed by humans.' is absent from the hero section where 80% of users decide to stay or leave.
**Fix:** Add a line to the hero body copy: 'When you get stuck, our team jumps in to help. Free white-glove onboarding included.' to match the meta description promise.
### Copy Specificity & Tone
**Score:** 6/10
**Issue:** The phrase 'It's the complete system' is generic and could apply to an ERP, a CRM, or a CMS. The feature chips like 'Audit Logs', 'SSO', 'SQL Integrations' are feature-led labels without context on why they matter to a non-technical founder.
**Fix:** Rewrite the subheadline to 'The internal tool builder that ships with security, hosting, and support.' Rewrite feature chips to benefit-led text like 'Enterprise Security' instead of 'SSO'.
### Page Narrative & Flow
**Score:** 5/10
**Issue:** Section 1 displays metrics like '+67% Process Efficiency' before Section 2 asks 'Sound familiar?' to identify pain points. Visitors see the gain before feeling the problem.
**Fix:** Move the 'Sound familiar?' section to immediately follow the Hero. Place the 'On average, Agent UI customers achieve' metrics after the problem agitation to validate the solution.
### Anxiety & Objection Handling
**Score:** 4/10
**Issue:** The reassurance 'No credit card. Start in minutes.' appears in Section 10, far below the Hero CTAs. Pricing is only accessible via the navigation link 'Plans & Pricing', not in the page body.
**Fix:** Use only verified page facts here: Plans start at [actual starting price] | [actual trial or credit-card policy] | [actual cancellation, setup, or security detail]. If the site does not state a fact, ask the founder to add the real detail instead of inventing it.
### CTA Architecture
**Score:** 3/10
**Issue:** Hero section uses 'Build your first app' (dark button) and 'Book a Call' (white button), but section 10 uses 'Create my system free' and footer uses 'Start Building Free'. Scraped data indicates the hero 'Book a Call' button has a missing href destination.
**Fix:** Standardize all self-serve CTAs to 'Start Building Free'. Ensure every button has a valid href. Make 'Build your first app' the single primary action.
### Friction & Distraction
**Score:** 4/10
**Issue:** Footer contains 12 language switcher buttons (EN, ES, PT, etc.) plus 'Join our Discord', social links, and 6+ Trustpilot review links. Hero pits high-friction 'Book a Call' against low-friction 'Build your first app'.
**Fix:** Collapse language selector into a single dropdown. Remove Discord and social links from the conversion path. Deprioritize 'Book a Call' below the fold for PLG focus.
### Trust & Social Proof
**Score:** 5/10
**Issue:** The hero section relies on the vague claim 'Trusted by companies in 15 countries' instead of named logos. Strong testimonials like 'Nazareno Delgado, Entrepreneur - Argentina' are buried in section 7, far from the primary 'Book a Call' CTA.
**Fix:** Move a 3-logo customer strip or the 'Nazareno Delgado' testimonial snippet directly under the H1 'It's not a demo.' Replace '15 countries' with specific industry names or recognizable brands if available.
### Visual Hierarchy
**Score:** 4/10
**Issue:** Mobile sticky overlays cover 72 percent of the viewport obscuring the It's not a demo. headline, and 48 tap targets including Build your first app are below the 48px accessibility standard.
**Fix:** Reduce sticky header and cookie banner height to under 15 percent of viewport, increase Build your first app and Book a Call hit areas to 44px minimum, and compress hero padding on mobile to surface +67% Process Efficiency metrics above the fold.
### Color & Accessibility
**Score:** 3/10
**Issue:** Lab data flags 1 contrast risk likely in the purple subhead It's the complete system. and 48 small tap targets including footer language buttons like EN and feature pills like Audit Logs plus 1 input missing a label.
**Fix:** Darken purple text to meet WCAG AA 4.5:1 ratio, increase all button and link padding to minimum 44x44 pixels, and add visible labels or aria-labels to all form inputs.
### Mobile Experience
**Score:** 3/10
**Issue:** Mobile lab data shows 48 small tap targets below the 44px minimum, a hero_height_ratio of 23.28 indicating extreme vertical bloat, and a sticky_overlay_area_ratio of 0.72 covering most of the screen.
**Fix:** Increase all button and link hit areas to minimum 44x44 CSS pixels, reduce hero section padding to fit within 100vh on mobile, and shrink sticky overlay to max 15% viewport height.
### SEO Fundamentals
**Score:** 3/10
**Issue:** Canonical URL is null, sitemap excludes homepage and 48 crawled URLs, 6 pages share identical title "AgentUI - AI + Human App Builder | Build Apps with Human Backup", and broken link "https://www.agentui.ai/enmailto:support@agentui.ai" exists. H1 "It's not a demo." lacks target keyword.
**Fix:** Use only verified page facts here: Plans start at [actual starting price] | [actual trial or credit-card policy] | [actual cancellation, setup, or security detail]. If the site does not state a fact, ask the founder to add the real detail instead of inventing it.
### Load Speed
**Score:** 3/10
**Issue:** High volume of third-party requests (56 total) with a single domain, app.termly.io, making 22 requests.
**Fix:** Audit the Termly implementation to reduce request count or replace it with a more lightweight consent management provider.
### Technical Health
**Score:** 4/10
**Issue:** The page contains a broken internal link to 'https://www.agentui.ai/enmailto:support@agentui.ai' (404) and exhibits severe metadata duplication, with 6 different URLs sharing the exact same title 'AgentUI - AI + Human App Builder | Build Apps with Human Backup' and meta description.
**Fix:** 1. Correct the malformed email link to a standard 'mailto:support@agentui.ai' format. 2. Create unique, keyword-optimized titles and descriptions for the affected pages: /en/build-faster/, /en/affiliates/, /en/cookies/, /en/terms/, and /en/dsar/. 3. Update the sitemap to include all 50+ crawled pages instead of the current single-entry sitemap.


## Validation Checklist

- `performance.blocking_scripts`: Re-run the audit and confirm this finding is resolved: Render-blocking scripts detected.
- `performance.image_dimensions`: Re-run the audit and confirm this finding is resolved: Images are missing reserved dimensions.
- `performance.request_count`: Re-run the audit and confirm this finding is resolved: Too many network requests.
- `performance.script_count`: Re-run the audit and confirm this finding is resolved: JavaScript file count is high.
- `performance.third_party_domain_count`: Re-run the audit and confirm this finding is resolved: Too many third-party origins.
- `performance.third_party_requests`: Re-run the audit and confirm this finding is resolved: Too many third-party requests.
- `accessibility.form_labels`: Re-run the audit and confirm this finding is resolved: Form controls lack labels.
- `visual.sticky_overlay_blocks_mobile`: Re-run the audit and confirm this finding is resolved: Sticky overlay takes too much mobile space.
- `cro.primary_cta_no_destination`: Re-run the audit and confirm this finding is resolved: Primary CTA has no inspectable destination.
- `visual.obvious_overlap`: Re-run the audit and confirm this finding is resolved: Potential element overlap detected.

## Output I Want From You

- Make the code changes directly.
- Show a concise summary of what changed.
- List any placeholders I need to replace with real business facts.
- Mention which checks you ran.