UI Discovery

UI Discovery is a design-reference catalog, not a design tool: 2,647 websites, 62 web apps and 3,836 iOS apps (https://uidiscovery.com/about), captured and cut into 111,083 individually labelled sections — heroes, pricing tables, footers, onboarding screens. Each capture ships with an extracted colour palette, font stack, headings and an AI-ready 'build prompt'. Website and web-app captures come from the vendor's own crawler screenshotting public pages (no logins or paywalls bypassed, per its sourcing page); iOS screens come from public App Store listings, the vendor's own in-app recordings, or — for some app collections — imported third-party reference archives, which the product labels as such on the relevant app pages (https://uidiscovery.com/sources).

This is the tool's clearest strength. The hosted MCP endpoint (https://uidiscovery.com/mcp) exposes a specific, documented tool set — search_sections, search_screens, search_flows, search_sites, compare_references, find_similar, get_design_tokens, and Pro-gated get_design_brief / get_image_plan — plus a parallel REST API (/api/v1/sections, /sites, /search, /brief, /flows) with documented filters, pagination (25 records/request) and rate limits (120 list reads/minute per connection) (https://uidiscovery.com/api-docs). The product's stated workflow is deliberate: search → inspect real references → produce an original artifact with new copy, the buyer's brand, and royalty-free or generated imagery, never the reference's own assets — a position repeated in both the terms of use and the MCP system prompt (https://uidiscovery.com/terms, https://uidiscovery.com/connect). For a team building 'make me a site like X' tooling, that's a cleaner API contract than most reference libraries publish.

Free: browsing the full library, 3 full designs (prompt + code) a day, and 200 API requests/day. Pro: $15/month or $150/year (two months free), unlimited designs, high-res downloads, and 5,000 API requests/day. Team: $39/month, adds 50,000 API requests/day on a single account — the page is explicit that shared workspaces and extra seats are 'not available yet' despite the Team name. All three tiers are listed with real numbers on the public pricing page, billed through Stripe (https://uidiscovery.com/pricing).

The things the site says about itself don't agree. The careers page claims 'a team of 25 designers, developers and engineers' with 16 open remote roles across design, engineering, product and growth — but applications go through a plain mailto link, not an applicant-tracking system (https://uidiscovery.com/careers). Both the terms of use and the privacy policy state plainly: 'UI Discovery is operated by Ramandeep Singh ("UI Discovery", "we", "us")' — a single named individual, not a company entity — governed by the laws of Alberta, Canada, with support handled from a Gmail address ([email protected]) rather than a company domain (https://uidiscovery.com/terms, https://uidiscovery.com/privacy). Domain registration (RDAP, Hostinger) shows uidiscovery.com was registered on 2026-09-24, and the Internet Archive's earliest capture of the live site is 2026-10-10 — meaning the product is roughly three weeks old at the time of this review. None of that makes the library fake (it works, and the API is real), but a buyer evaluating it as infrastructure for a product should treat the 'team of 25' framing as unverified marketing copy, not as a signal of institutional maturity.

The vendor is upfront that captured designs, logos and screenshots belong to their original owners, that the library exists for study and reference rather than redistribution, and that owners can request removal of specific references or an entire product by email (https://uidiscovery.com/content-removal). That's a reasonable, if informal, takedown process for a reference library. Two of the policy pages that carry this commitment — Sources & Rights and Content Removal — are each marked 'Draft, last updated September 29, 2026' directly on the page, meaning the legal terms a rights holder or buyer would rely on are explicitly still being finalized.

The privacy policy is unusually specific for a tool this young: search and design-brief text sent to its AI ranking provider (TypeSafe) excludes any account or device identifier, per-IP abuse counters live only in server memory and are never persisted, daily visitor counts are computed from a same-day salted hash rather than stored IPs, and the vendor states it does not sell personal data or use it for advertising (https://uidiscovery.com/privacy). No SOC 2, DPA or other formal compliance documentation is published, and there is no separate security/trust page — this is a plain-language privacy policy, not a compliance program.