Monitor ChatGPT and Claude for Mentions of Your Logistics Company

Marketing teams at logistics companies rely on tools like SEMrush or Ahrefs for visibility. These platforms are excellent for tracking Google rankings and backlinks, but they cannot see what happens inside an AI chat session. They are built to crawl the public web, but AI-generated answers are not public web pages. This creates a massive blind spot in your marketing intelligence.

Other teams might try brand monitoring tools like Google Alerts or Meltwater. These services scrape news sites, blogs, and social media for company mentions. The failure mode is simple: they see nothing from AI chats. A potential shipper could ask Claude, 'Who are the best logistics partners for e-commerce returns in Illinois and Ohio?' and get a recommendation for your company. Your team would have no idea this discovery path even exists because the conversation is private and inaccessible to web scrapers.

Here is a concrete scenario. Your marketing manager wants to know if you are being recommended for 'last-mile delivery solutions in the Midwest.' She sets up her dashboards and sees your Google rankings. Meanwhile, a retail operations director asks ChatGPT a series of increasingly specific questions about her supply chain problems, landing on your company because you have a detailed case study about tile-industry-specific warehousing. She books a call. You have a new lead, but you are completely blind to how it arrived.

The structural problem is that AI chat responses are generated on-demand within a user's private session. They are not static, indexable URLs that a traditional crawler can find. The only way to see what an AI is saying about you is to actively query its API at scale, simulating hundreds of different buyer questions. Off-the-shelf monitoring tools are not architected for this query-based model.

We built our own system by first identifying the key questions a buyer would ask. These are not just keywords like 'freight forwarding' but full-sentence problems like 'how can I reduce demurrage fees at the Port of Long Beach?'. We built a list of over 50 such prompts covering different service lines and problems relevant to our ideal client.

The core of the system is a Python script that runs on a weekly schedule using AWS Lambda. The script iterates through our prompt list and sends them to the APIs of 9 different AI models, including OpenAI's GPT-4, Anthropic's Claude 3, and Google's Gemini. We use the httpx library for parallel, asynchronous calls to speed up the process. The system parses each text response, searching for mentions of 'Syntora' or our URL, and writes all findings to a Supabase database table.

The result is a dashboard that shows our 'Share of Voice' across the AI ecosystem. We see which prompts trigger a recommendation and which models cite us most often. This data directly informs our content strategy. For a logistics company, a custom version of this system would provide a direct view into how and when AI recommends your specific services, from drayage to cold chain storage.