getting company logo from crawl

hi folks, not having much success retrieving a logo url for a brand (e.g. apple.com) with the crawl endpoint

my current process is
to crawl the url via firecrawl api, and then send the html and markdown to an llm, asking the llm to find (among other things) the url of the logo (using a zod schema) from the html / markdown

here's our firecrawl code:

    const crawlResponse = await firecrawl_app.scrapeUrl(url, {
      formats: [
        "markdown",
        "html",
        ...(options.includeScreenshot ? (["screenshot@fullPage"] as any) : []),
      ],
      ...(options.includeScreenshot
        ? {
            actions: [{ type: "screenshot", fullPage: true }],
          }
        : {}),
    });

    const crawlResponse = await firecrawl_app.scrapeUrl(url, {
      formats: [
        "markdown",
        "html",
        ...(options.includeScreenshot ? (["screenshot@fullPage"] as any) : []),
      ],
      ...(options.includeScreenshot
        ? {
            actions: [{ type: "screenshot", fullPage: true }],
          }
        : {}),
    });

we then pass this result to an llm, and here's an example result from the llm for Anthropic (you can see it gets the logo url wrong):

brand: {
    logoUrl: 'https://www.anthropic.com/jobs',
    logoUrlEvidence: "The Anthropic logo is in the bottom left corner of the webpage. It is a simple 'A' symbol.",
    name: 'Anthropic',
}

brand: {
    logoUrl: 'https://www.anthropic.com/jobs',
    logoUrlEvidence: "The Anthropic logo is in the bottom left corner of the webpage. It is a simple 'A' symbol.",
    name: 'Anthropic',
}

any tips on getting the logo url successfully after a scrape?

do you recommend a different endpoint? We like that scrapeUrl seems fast

getting company logo from crawl

    const crawlResponse = await firecrawl_app.scrapeUrl(url, {
      formats: [
        "markdown",
        "html",
        ...(options.includeScreenshot ? (["screenshot@fullPage"] as any) : []),
      ],
      ...(options.includeScreenshot
        ? {
            actions: [{ type: "screenshot", fullPage: true }],
          }
        : {}),
    });

    const crawlResponse = await firecrawl_app.scrapeUrl(url, {
      formats: [
        "markdown",
        "html",
        ...(options.includeScreenshot ? (["screenshot@fullPage"] as any) : []),
      ],
      ...(options.includeScreenshot
        ? {
            actions: [{ type: "screenshot", fullPage: true }],
          }
        : {}),
    });

we then pass this result to an llm, and here's an example result from the llm for Anthropic (you can see it gets the logo url wrong):

brand: {
    logoUrl: 'https://www.anthropic.com/jobs',
    logoUrlEvidence: "The Anthropic logo is in the bottom left corner of the webpage. It is a simple 'A' symbol.",
    name: 'Anthropic',
}

brand: {
    logoUrl: 'https://www.anthropic.com/jobs',
    logoUrlEvidence: "The Anthropic logo is in the bottom left corner of the webpage. It is a simple 'A' symbol.",
    name: 'Anthropic',
}

any tips on getting the logo url successfully after a scrape?

do you recommend a different endpoint? We like that scrapeUrl seems fast

getting company logo from crawl

Similar Threads

getting company logo from crawl

Similar Threads

Similar Threads

Similar Threads