How AI assistants recommend industrial equipment: gantry and davit cranes - ON-1267

Project type: Research
Desired discipline(s): Computer science, Mathematical Sciences, Statistics / Actuarial sciences, Business, Social Sciences & Humanities
Company: Easily Moved Equipment ULC (eme)
Project Length: 4 to 6 months
Preferred start date: 01/04/2027
Language requirement: English
Location(s): ON, Canada
No. of positions: 1
Desired education level: Master'sPhD
Open to applicants registered at an institution outside of Canada: No

About the company: 

Easily Moved Equipment ULC (eme) has designed and built portable aluminum lifting equipment in Ontario, Canada since 2003, and sells it across Canada and the United States. Our two primary product lines are portable aluminum gantry cranes, rated from 1,100 to 22,000 lb and designed to roll while fully loaded, and socket-mounted aluminum davit cranes, where one arm serves a network of sockets to lift pumps and equipment at water and wastewater facilities. We also make aluminum lifting beams and accessories.

We sell through equipment rental companies, industrial distributors, mechanical and general contractors, municipalities and industrial end users.

We are a team of six, so we rely on software. We run quoting and operations on a platform we built in-house. In 2026 we relaunched easilymovedequipment.com in English, French and Spanish, with machine-readable product data (llms.txt, product feeds, structured data) for AI crawlers. What we lack is evidence: we do not know how AI assistants decide which lifting equipment to recommend, or which of our efforts change that decision.

Describe the project.: 

Industrial buyers now ask AI assistants (ChatGPT, Google AI Overviews and AI Mode, Gemini, Copilot, Perplexity, Claude) which equipment to buy and from whom. The assistant answers with a short list of brands and a few cited sources. A manufacturer left off that list never learns the buyer existed. ChatGPT already sends visitors to our website, but early measurements show larger competitors named far more often. How these systems choose what to recommend is poorly understood, especially in small, technical B2B categories.

eme's goal is actionable knowledge: an evidence-based model of how generative AI recommends products in our two primary categories, portable aluminum gantry cranes and davit cranes, and a tested playbook for designing our website and digital footprint around it.

The intern will:

  1. Build a prompt set that mirrors real buyer questions by role (rental branch, contractor, plant engineer, utility maintenance lead, municipal wastewater operator), application, capacity and market (Canada, US), using our search-console and sales-inquiry data.
  2. Run the prompts repeatedly across the major Record which brands each recommends, in what order, with which claims, and which pages it cites. Quantify run-to-run and model-to-model variance.
  3. Analyze the cited sources (manufacturer sites, distributors, rental catalogues, directories, forums, video) and the page features linked to being cited or recommended: structured data, specification tables, comparison and FAQ content, third-party mentions, freshness and crawler access.
  4. Test changes to our site and off-site presence, and measure their effect against the baseline and a competitor control set.
  5. Deliver a reusable measurement tool, a findings report and a prioritized roadmap for our website and digital footprint.

Methods: controlled prompt experiments with repeated sampling, citation analysis, statistical modelling of recommendation outcomes, and before/after intervention studies. The findings will help any Canadian manufacturer of niche technical products.

Required expertise/skills: 

  • Required:
  • Python for data collection and analysis (pandas, Jupyter, HTTP APIs).
  • Working knowledge of large language models and retrieval-augmented generation: how AI assistants search the web and choose sources. Experience calling LLM APIs (OpenAI, Google Gemini, Anthropic) and handling non-deterministic output.
  • Experimental design and applied statistics: repeated sampling, variance, significance testing, regression or classification.
  • Clear writing for a business audience.

Assets:

  • Search engine optimization and web analytics (Google Search Console, Bing Webmaster Tools, Google Analytics, Ahrefs or Semrush). Structured data (schema.org JSON-LD), HTML and static-site frameworks (Astro).
  • Web scraping, information retrieval or NLP (entity extraction, embeddings).
  • Data visualization (Python plotting, Looker Studio or Power BI).
  • Interest in B2B marketing, industrial equipment or workplace safety.
  • French, for Quebec-market prompts.

Express interest