Emerged

The Problem

Product discovery is fragmented and manual

Discovering new products means manually checking Product Hunt, Hacker News, Reddit, Indie Hackers, BetaList, and more — every day. There's no single place that curates across sources, personalises to your interests, and surfaces what actually matters.

I designed and built Emerged to solve this — a product discovery platform that automatically collects from 8 sources, uses AI to score, categorise, and curate, then delivers a personalised experience. I led every design decision from product strategy and information architecture to visual design and component patterns. AI (Claude) handled the code implementation under my direction. A platform of this scope — 46 pages, 105 components, 3 AI models, full monetisation — would normally require a team of engineers and months of runway. I shipped it solo, in weeks. Emerged is production-ready and launching soon.

The Design Scope

A complete SaaS product designed end-to-end

Emerged serves three audiences — end users discovering products, founders promoting their listings, and admins moderating at scale. Each required a distinct design approach while maintaining a coherent product. The platform includes onboarding and personalisation flows, a custom design system, monetisation (sponsorships, founder perks, launch packages), a full email system, and production monitoring.

Designing for Users

Personalised discovery without effort

The feed: A YouTube-style recommendation engine

The core experience is a personalised feed powered by a recommendation algorithm comparable in complexity to YouTube's approach. Each user has a taste profile — built from category affinities, dimension preferences (audience, platform, industry, technology), interaction history (saves, upvotes, impressions, view duration), and collaborative filtering against similar users. The algorithm weights recency, engagement signals, trending scores, and editorial quality to produce a relevance-ranked feed that improves with every interaction.

Taste profiles are recomputed nightly. Feeds are pre-cached every 30 minutes for instant loading. Trending scores recalculate every 2 hours. The result is a feed that feels effortless to the user but is backed by sophisticated personalisation — the kind of system design challenge typically reserved for teams of engineers.

The design challenge was making this personalisation feel transparent, not opaque. Users can see their taste profile in settings, fine-tune it by excluding categories or dimensions, and understand why products appear. A “while you were away” banner re-engages returning users by surfacing what they missed.

Product details: Staying in the flow

I designed an intercepting route modal for product details — clicking a card opens a detail panel over the feed. Users can read the AI editorial take, view screenshots, see similar products, save to collections, and share — all without losing their scroll position. Direct URLs fall back to a full standalone page, supporting deep-linking and social sharing.

Transferable skill: The browse-then-detail pattern (inspired by Instagram and Product Hunt) is applicable to any content-heavy product — e-commerce, job boards, media platforms. The key decision is when to break the browsing flow and when to preserve it.

Collections: User-curated and AI-generated

Users can save products to collections manually, or describe what they want (“tools for indie hackers building in public”) and get an AI-curated list generated automatically. Collections have auto-generated cover artwork, can be reordered, and can be shared publicly with unique URLs.

Designing for Founders

Claim, customise, and promote — a second product inside the product

Founders are a distinct user type with different goals. They don't want to discover products — they want their product discovered. I designed a founder experience inside Emerged: claim your listing through email verification, then customise it with hero images, demo videos, custom CTAs, and discount perks.

Edit requests go through an approval queue to maintain content quality — a design pattern that balances founder autonomy with editorial control. A “My Products” dashboard shows claimed and submitted products with their current status.

Monetisation design

I designed the monetisation model from scratch: individual perks (verified badge, video embed, custom CTA, custom hero) and all-in-one launch packages that bundle perks with featured placement at a discount. Sponsorships run for 1–2 weeks with impression tracking. The pricing structure was designed to make the bundle feel like obvious value — individual perks total $17.96, the full launch package is $11.99.

Transferable skill: Designing monetisation flows — pricing psychology, upgrade paths, feature gating, and payment integration — is a core product design skill for any SaaS or marketplace product.

Designing for Admins

Moderation at scale with AI-assisted triage

Hundreds of products are scraped daily from 8 sources. Without smart moderation, an admin would spend hours reviewing every listing. I designed a 6-stage approval pipeline that uses AI to handle the volume:

1. Products enter as Pending Triage
2. AI scores quality (0–100 with confidence) — high-confidence rejections auto-skip, removing the biggest time sink
3. Admin reviews remaining as Interested or Passed
4. Approved products move to Ready to Queue
5. The publishing queue controls timing and pacing
6. Published — live on the platform, founder notified

The system learns from admin decisions over time — when an admin overrides an AI triage score, that feedback improves future scoring. This human-in-the-loop pattern reduces admin workload without sacrificing quality.

Transferable skill: Designing AI-assisted moderation workflows — where AI handles volume and humans handle judgment — is directly relevant to content platforms, marketplaces, and any product with user-generated content.

Beyond moderation, the admin panel includes email campaign management with A/B testing and delivery tracking, cultural phase management (AI-detected trends in the startup landscape), engagement analytics, scraper monitoring with job history, and a full audit log.

My Approach to AI Product Design

Four patterns I apply to every AI-powered feature

Building two AI-powered products has taught me that every AI feature needs the same four design patterns. These aren't specific to Emerged — they're the framework I bring to any AI product work:

As every product adds AI capabilities, these patterns — fallback chains, confidence gating, AI-as-draft, and model selection — are becoming core product design competencies. I've applied them across two production platforms and can bring this framework to any team building with AI.

Design System

A 105-component library with a custom glass design language

With 46 pages and three distinct user types, I needed a design system that scaled. I designed a two-tier elevation approach: grounded surfaces (cards, panels, backgrounds) use subtle shadows, while floating elements (modals, dropdowns, tooltips) use backdrop blur and inset highlights to create depth — a glass design language.

The 105-component library is organised by domain — startup cards, feed components, founder tools, admin interfaces, browse filters, lists, and shared UI primitives. Each component handles its own loading, error, and empty states. The system is documented in a design system reference covering elevation classes, colour tokens, shadow scale, z-index hierarchy, responsive breakpoints, and padding standards.

Email & Engagement Design

Bringing users back with personalised communication

Discovery platforms die without re-engagement. I designed the full email experience: personalised daily and weekly digests, founder approval notifications written by Claude (mentioning where the product was discovered and what stood out), follow-up emails for inactive users, and a CAN-SPAM compliant unsubscribe flow.

The admin email tools support A/B testing, delivery tracking (sent, delivered, bounced, complained), rate limiting, and template management. Email drafts move through a status pipeline (draft → approved → scheduled → sent) with an AI safety mode that prevents accidental sends during testing.

Hi there,

Great news — Luminai has been published on Emerged! We discovered it on Product Hunt and were impressed by how it bridges design tokens and production code.

Your listing is live and already appearing in personalised feeds for users interested in design tools.

Claim your listing to customise it with a hero image, video demo, and custom CTA.

Best,
The Emerged Team

What I Bring

A designer who understands AI from both sides

I design AI-powered features — personalisation, moderation, scoring, recommendations — and I use AI to build production systems. That means I understand AI as a product feature (what to show users, when to involve humans, how to handle failure) and as a design tool (how to move from concept to shipped product at a pace that wasn't possible before).

Here's what I'd bring to your team:

• An AI design framework — four tested patterns (fallback chains, confidence gating, AI-as-draft, model selection) that I can apply to any AI feature your product is building. Not theory — patterns proven across two production platforms.
• Multi-audience product thinking — I've designed for consumers, creators, and admins within one product. The same challenge every marketplace, platform, and B2B tool faces.
• End-to-end product design — from market gap identification to monetisation strategy to 105-component design system. I think in products, not just screens.
• Speed that changes what's possible — I shipped a 46-page platform with 3 AI models, full monetisation, and production infrastructure as a solo designer-founder. When your team needs to move fast on an AI initiative, I can prototype, validate, and ship in a fraction of the time.