public.com — AI Site Grade

Analysis

Public.com — AI-Visibility Audit

The cold LLM knowledge of Public.com is stuck in 2021–2023 (social trading, anti-Robinhood, no-PFOF, "Public Vaults," layoffs), while the live site has fully pivoted to an "agentic brokerage" identity — AI Agents, Generated Assets, API trading, options rebates, and direct indexing — that the model has almost no awareness of.

Crawler Access

Every major AI crawler — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, anthropic-ai, Bytespider, Applebot-Extended — receives a full 200 with identical byte payload (232KB) on the homepage. No UA-based blocking, no Cloudflare challenge, no JS shell. The robots.txt is a bare WordPress skeleton: no AI-bot directives at all, just a Disallow: /wp-admin/ for the wildcard rule. The llms.txt exists (generated by Yoast SEO v27.6) and lists 5 pages, 5 learn articles, 5 tools, and some category/tag links — a decent start but thin for a site with hundreds of indexed pages. The sitemap_index.xml is well-structured with sub-sitemaps for pages, learn articles, tools, stock tickers, and "how-to-buy" content. The API docs page (/api/docs) is JS-rendered — bots get a 200 with 62KB but extract 0 words of visible text, meaning AI crawlers cannot read the API documentation.

Cold-Knowledge Gap

The frontier model describes Public as a "commission-free platform with social features" that "gained prominence during the 2021 GameStop meme stock frenzy" and mentions "Public Vaults," "Public Premium," and "scrutiny over interest rate disclosures." This is entirely backward-looking. The live site makes zero mention of social features, tipping, or the GameStop era. The homepage headline is "Investing for those who take it seriously" and leads with AI Agents ("world's first agentic brokerage"), Generated Assets (AI-created custom indexes), options rebates, margin at 4.90%, and API programmatic trading. The model knows nothing about the Agentic Brokerage launched November 2025, the Alto crypto IRA acquisition ($65M), the direct indexing launch, or the options rebate program. The gap is extreme: the model would describe a social-trading app that no longer exists.

Schema Posture

The homepage and /about-us pages carry a single Organization schema block with founders, address, founding date, and 13 sameAs links (Wikipedia, LinkedIn, Crunchbase, app stores, socials). This is solid but static — there is no WebSite schema, no SearchAction, no FAQPage schema despite FAQ sections on multiple pages, no Product or SoftwareApplication schema for the app, and no BreadcrumbList on most pages (only /learn has one). Key product pages (/ai-agents, /generated-asset, /invest/options-trading) have zero JSON-LD schema at all.

External Signals

The Wikipedia article (last modified April 2026) is the strongest external signal and is well-aligned with the current site — it documents the Agentic Brokerage launch (November 2025), the Alto acquisition, direct indexing, and options rebates. The Wikipedia page carries a neutrality-dispute banner (April 2026) noting a "close connection with its subject." DNS TXT records confirm domain verification for OpenAI (openai-domain-verification=dv-tcRgG9s1INKuEPpU629VwhRG), Anthropic (anthropic-domain-verification-15srwp), Google, Apple, Facebook, Atlassian, Stripe, and Adobe — indicating active engagement with AI platform ecosystems. No recent press or Reddit threads surfaced in search results.

Notable Surprises

The /api/docs page is invisible to AI crawlers despite being fully accessible — it renders via JavaScript and returns zero extracted text. This is the documentation for the very API that the site markets as a key differentiator. The /research/2024-retail-investor-report page is also a thin JS shell (43 words extracted). The learn sitemap contains hundreds of articles, some dating back to 2020 (meme stock coverage, stale IPO guides), alongside fresh 2026 content — a mixed freshness profile. The llms.txt references a sitemap at hellopublic.wpengine.com (the WordPress engine hostname) rather than the canonical domain, a minor but sloppy detail.

Findings

1. Cold LLM knowledge describes outdated social-trading platform High

   Frontier models describe Public.com as a commission-free social-trading platform from the GameStop era, with no awareness of the 2025 Agentic Brokerage launch, AI Agents, Generated Assets, options rebates, or API trading. The live site has fully pivoted to an 'agentic brokerage' identity that the model has almost no knowledge of.

   What to change: Publish structured data (e.g., SoftwareApplication, Product) and press releases that clearly describe the current platform features. Ensure Wikipedia and other knowledge bases are updated with the latest offerings.

2. API documentation page is invisible to AI crawlers High

   The /api/docs page returns a 200 status but renders content via JavaScript, resulting in zero extracted words for AI crawlers. This means the documentation for the API that the site markets as a key differentiator is completely invisible to LLMs.

   What to change: Implement server-side rendering (SSR) or static HTML generation for the API docs page so that crawlers can extract the full text content.

3. Key product pages lack any JSON-LD schema High

   Pages such as /ai-agents, /generated-asset, and /invest/options-trading have zero JSON-LD schema markup. This limits the ability of AI crawlers to understand and cite these pages as authoritative sources for the platform's features.

   What to change: Add appropriate JSON-LD schema (e.g., SoftwareApplication, Product, FAQPage) to each product page to describe the features and functionality.

4. llms.txt is thin and references internal hostname Medium

   The llms.txt file lists only 5 pages, 5 learn articles, and 5 tools, which is sparse for a site with hundreds of indexed pages. Additionally, it references a sitemap at hellopublic.wpengine.com instead of the canonical public.com domain.

   What to change: Expand the llms.txt to include all major sections of the site (e.g., learn articles, product pages, research reports) and fix the sitemap URL to use the canonical domain.

5. Homepage lacks WebSite schema with SearchAction Medium

   The homepage only has Organization schema and no WebSite schema, which means AI crawlers cannot easily identify the site's search functionality or navigation structure.

   What to change: Add WebSite schema with SearchAction to the homepage to help crawlers understand the site's search capabilities.

6. Research report page is a thin JS shell Medium

   The /research/2024-retail-investor-report page returns only 43 words of extracted text, indicating it is rendered via JavaScript and largely invisible to AI crawlers.

   What to change: Implement SSR or static HTML for research report pages to ensure full content is accessible to crawlers.

7. Learn sitemap contains stale content alongside fresh articles Low

   The learn sitemap includes hundreds of articles, some dating back to 2020 (e.g., meme stock coverage, stale IPO guides), alongside fresh 2026 content. This mixed freshness profile may dilute the site's perceived authority on current topics.

   What to change: Consider removing or updating outdated articles to maintain a consistent freshness signal for AI crawlers.

8. FAQ sections on product pages lack FAQPage schema Low

   Multiple pages contain FAQ sections but do not use FAQPage schema, missing an opportunity to have answers directly surfaced in AI responses.

   What to change: Add FAQPage schema to pages with FAQ sections to enable rich results and direct answer extraction.

What's working

• All major AI crawlers receive full access to homepage — Every major AI crawler (GPTBot, ClaudeBot, PerplexityBot, etc.) receives a 200 response with identical byte payload on the homepage, with no UA-based blocking or JS shell.
• Domain verified for OpenAI and Anthropic — DNS TXT records confirm domain verification for OpenAI and Anthropic, indicating active engagement with AI platform ecosystems.
• Sitemap index is well-structured with sub-sitemaps — The sitemap_index.xml organizes content into sub-sitemaps for pages, learn articles, tools, stock tickers, and how-to-buy content, aiding crawler discovery.
• Homepage includes Organization schema with sameAs links — The homepage and /about-us pages carry a single Organization schema block with founders, address, founding date, and 13 sameAs links (Wikipedia, LinkedIn, Crunchbase, etc.), providing a solid foundation for entity recognition.
• Wikipedia article is well-aligned with current site — The Wikipedia article (last modified April 2026) documents the Agentic Brokerage launch, Alto acquisition, direct indexing, and options rebates, providing strong external signal alignment.
• llms.txt file is present and lists key resources — An llms.txt file exists (generated by Yoast SEO) and lists 5 pages, 5 learn articles, and 5 tools, providing a starting point for AI crawlers to discover important content.
• API docs page is accessible to bots despite JS rendering — The /api/docs page returns a 200 status and is accessible to all 11 tested bots, even though the content is not extractable due to JS rendering.