The agent retrieved only the meta description from drapper.com's homepage, which stated the company builds bespoke web and business apps for UK SMEs. Unable to access detailed pricing, service differentiation, or competitive positioning information, the agent fell back on web search (which returned no substantive results about Drapper itself) and inference. The site's React single-page architecture meant the HTML skeleton contained no rendered content, and the agent could not fully satisfy the task without JavaScript execution or direct contact.

• ›Step [1] (homepage fetch) was the only content-bearing step and provided only the meta description: 'Drapper Software builds bespoke web and business apps for UK SMEs — internal tools, dashboards and portals, built by people who run real businesses.' All other fetches ([3], [4]) returned identical HTML skeletons with no additional content.
• ›The agent attempted to navigate to /pricing and /about pages ([3], [4]) but received the same React SPA shell both times, indicating the site does not pre-render content in the HTML—a critical accessibility gap for non-JavaScript crawlers.
• ›Web searches ([6], [8], [10], [12]) returned zero substantive results about Drapper specifically; results instead surfaced competitors like DrapCode, Procosoft, and generic bespoke software development guides. This suggests Drapper has minimal SEO presence or is too new/small to index.
• ›The agent inferred positioning ('built by people who run real businesses' implies entrepreneurial developer experience) but had no source material to answer pricing, differentiation, or sales process questions, making the final response heavily inference-based rather than sourced.
• ›The site is extremely agent-hostile: no public pricing page, no structured metadata beyond a generic description, no case studies or testimonials accessible in HTML, and no fallback content for non-JavaScript clients. This forces the agent to either require JavaScript rendering or conduct live contact.