The agent successfully assembled a comprehensive explanation of iM Global Partner by relying almost entirely on web search results and a PDF corporate brochure, since the primary website (imgp.com) was largely non-functional—multiple pages redirected to a maintenance notice or the homepage. The agent delivered clear answers on what the company does (boutique asset manager network), who it serves (institutional investors), how it differs (independence-preserving partnership model vs. mega-firms), and pricing structure (asset-based fees, not publicly listed). However, specific fee percentages, exact partner count, and current product details had to be inferred from fragments rather than extracted from the site itself.

• ›The imgp.com homepage and subpages (pricing, services, about) all returned identical HTML and did not expose differentiated content—the site appears to be either misconfigured or under maintenance, making it unsuitable as a discovery source.
• ›The agent recovered by pivoting to web search (steps 8, 9, 10, 13–16, 18), which surfaced LinkedIn, PitchBook, Morningstar, and a corporate PDF brochure—these secondary sources, not the primary website, contained the factual substance (boutique partnership model, fund offerings, multi-product distribution).
• ›The PDF brochure (step 17) was the only direct company artifact that loaded successfully; it was cited in the final response sources but the agent's response indicates incomplete parsing (no specific detail extracted from it in the final text, suggesting PDF extraction was unsuccessful or minimal).
• ›Fee information was explicitly noted as unavailable on the website—the agent correctly identified that exact percentages require SEC Form ADV filings or direct contact, reflecting standard industry opacity but also highlighting a gap in self-service disclosure.
• ›Partner list inconsistency (9 vs. 10 partners) and varying partner names across sources indicate the website does not publish an authoritative, current roster of affiliated firms.
• ›The site's non-functional state forced the agent to rely on 66% external sources (web search + prior knowledge) rather than 33% direct site crawl, making this a heavy-bridge recovery typical of sites under maintenance or poorly indexed.