Affluent Labs · Local website review · October 31, 2026
Final - Chatgpt
A separate version of the final design with direct service navigation and rewritten copy throughout.
Open Final - Chatgpt Design gallery · Milestone plan · Original final design
Delivery
The new site lives in website/overhaul/final-chatgpt/. The gallery has an additional “Final - Chatgpt” card linking to the site and this report. The original final/, its copy source, earlier design options and existing reports were preserved. No cloud publication or deployment was performed.
Milestones
- Audit and focused research: reviewed the five original pages through their shared content, templates, navigation and scripts. Delegated AI policy research to the configured
gpt-6-lunaresearch agent. Identified repeated copy, buried services, unsupported outcome language and an unclear email-form action. - Service names and navigation: added direct links to IT Services, Cybersecurity, AI Training, AI Policy, and Website Setup/Care in every header and footer. Kept Home and Contact. Added dedicated pages for the three previously grouped services.
- Sitewide copy: rewrote the home page, every service, calls to action, contact instructions and metadata. Kept a consistent first-person voice and concrete deliverables. Removed style announcements, repeated process summaries and decorative security-status claims.
- Implementation and verification: reused the final design’s cards, timelines, styling, contact handler and responsive-menu pattern. Added local build and verification files; results are recorded below.
What changed
| Page | Result |
|---|---|
| Home | Introduces the five services directly and explains how a client starts. Removed filler statistics and the “Plain language” claim. |
| AI Training | One-on-one and group coaching; practical tasks; tool installation and setup; security, privacy and permission settings; process improvements and less repetitive work. Preserved two 45-minute monthly calls, account ownership and between-call help for the ongoing individual service. |
| AI Policy for Organizations | Approved tools and acceptable uses; confidential information; prompt security and untrusted documents, pages and messages; human review; employee guidance; quarterly reviews. Describes coordination with legal and compliance advisers without claiming certification or guaranteed compliance. |
| Cybersecurity | New copy for external and internal penetration testing, vulnerability assessment and compliance readiness. Explains authorization, scope, testing windows, reporting, remediation priorities and agreed retesting. |
| IT Services | New copy for advice and projects, security platform setup, security software training, and configuration/support. Security software training remains distinct from AI Training. |
| Website Setup and Care | Clarifies the existing scope: technical setup for a finished design, hosting/domain configuration, email, forms, launch and monthly care. Replaces uptime and prompt-response promises with concrete work. |
| Contact | Lists the current services and explains that “Prepare email” opens a draft for the visitor to send. Preserves the existing mailto flow and adds a direct email link. |
Routes and preservation
Home, IT Services, Cybersecurity and Contact retain their filenames inside the separate version. New pages are ai-training.html, ai-policy.html and websites.html. The old more-services.html route remains as a compatibility directory, preserving #coaching, #policy and #websites anchors. “More Services” no longer appears in the new site’s navigation or customer-facing copy. Historical versions retain their original wording.
Verification results
PASS — no outstanding failures in the local checks.
- Preservation: SHA-256 hashes match for all 62 protected original files. The gallery comparison confirms the only change there is the new card.
- New site: eight pages, 72 page/viewport combinations, and 89 link/asset/anchor checks passed. Widths: 320, 375, 390, 768, 1024, 1120, 1121, 1280 and 1440 pixels.
- Navigation and layout: all five service links appear consistently; desktop navigation fits; mobile menus open by keyboard and close with Escape; mobile menu links meet the 44-pixel height check. No horizontal overflow or text-contrast failures were detected by the reused design audit.
- Content and accessibility checks: one main heading per page, unique descriptions, correct current-page indication, no outdated service names or template residue, visible keyboard focus, working no-JavaScript navigation, and reduced-motion content checked. Normal-motion scroll reveals also passed on every page.
- Browser behavior: no console errors, uncaught page errors or failed requests in the page matrix. Nonlocal requests were blocked. Service navigation, legacy anchors and browser form validation passed; the email-draft handler passed a separate simulation.
- Deliverables: gallery, plan and report returned HTTP 200, passed the mobile overflow check, and their local links resolved. Clicking the new gallery entry opened the correct version.
- Existing project suite:
npm testpassed for eight production pages and 429 local references. This is separate regression coverage; the new version is covered by its own check above. - Visual review: inspected desktop home, IT Services and Cybersecurity captures and mobile AI Training, AI Policy and Contact captures. Desktop and mobile screenshots for all eight pages are saved in
final-chatgpt/_qa/.
The first text scan falsely matched “NaN” inside “maintenance”; whole-word matching corrected the test. A long command in this report initially overflowed on mobile; it now wraps. The final complete run passed after both corrections.
Research used
The research informed the topics and service boundaries; all website copy was written for this version. Comparable providers describe their own offers and are not evidence of this business’s qualifications or results.
- NIST Generative AI Profile: risk-management topics, information security and prompt injection. A guidance framework, not a certification.
- UK government employee guidance: practical treatment of sensitive information and checking outputs. Its public-sector rules are not presented as rules for every client.
- Microsoft prompt-injection guidance: direct and indirect attacks, untrusted inputs and limited tool access.
- Operator Intelligence AI governance service: a comparable service covering tool inventories, acceptable-use policies, data rules and training.
- RSM AI governance service: policy development and implementation planning as defined consulting deliverables.
Repeatable local checks
Run from C:\Claude_Working\AI Concierge with the existing local server running:
node website/overhaul/final-chatgpt/build.mjs
node website/overhaul/final-chatgpt/check.mjsEdit final-chatgpt/content.mjs for this version’s copy and rebuild. Its source is independent of content-final.mjs and the original copy-lab winners. The check uses the existing Playwright installation and design-review contrast/focus audit. Evidence: browser check results.
Limits
Responsive sizes were tested in off-screen installed Chrome with a throwaway profile, not on physical phones or in the user’s Firefox. Email draft construction was simulated without opening an email app; browser form validation was checked separately. No email was sent and delivery was not tested. No security assessment or compliance audit of a client system was performed. This version received editorial and functional review, not a new Jev scoring run. There are no new certifications, prices, response-time commitments or custom-model services in the copy.