CORE DIRECTORY // SYSTEM.USER.DIANA_ISMAIL
Labs by Diana — Experiments that ship.
Side projects that got out of hand. AI tools built for problems I kept tripping over — now live, now yours.
Diana Ismail — Portfolio
MODULE_007
TECHNICAL_OVERVIEW
A ground-up migration from WordPress to Next.js 16, designed and built by Diana's AI team. Every project page includes an AI Context Panel — a collapsible sidebar on desktop, inline accordion on mobile — that streams a neutral, AI-generated summary of the work on demand.
The Digital Twin chat engine is embedded site-wide: a floating widget on every page, and the primary interface on the Ask Diana page. Conversation history syncs across both via Redis, so a visitor can start in the widget and continue on the full page without losing context.
The site uses locally hosted fonts to avoid build-time network dependencies, WordPress URL redirects to preserve old bookmarks, and WCAG AA-compliant theming in both light and dark mode. All content is managed through Sanity Studio — the Owner updates copy, projects, and metadata without code changes. Deployed on Railway with semantic-release for automated versioning.
PROJECT_LEARNINGS_LOG
KEY_LEARNING_01
Turbopack on Railway cannot reach Google Fonts during production builds — switching to local woff2 files via next/font/local eliminated the network dependency entirely.
KEY_LEARNING_02
The parent nav's backdrop-blur creates a CSS stacking context that traps position: fixed children — a portal to document.body was the clean escape.
KEY_LEARNING_03
The AI Context Panel caches generated summaries in sessionStorage keyed by content hash — revisits within the same session skip the LLM call entirely, reducing token usage and latency without requiring backend state.