Skip to content

recommendation-engine

Target Audience: Consultants and operators building advisory MVPs that map a client or lead profile to a ranked shortlist of options (insurance plans, AI tools, vendors, suppliers, methodologies). Typical use is a sales-advisory demo, a discovery proposal for a new lead, or an internal decision-support tool for an existing client.

  • Area: Sales & Clients
  • Category: sales
  • Level: advanced
  • Input: conversation
  • Output: document
  • Setup: install
  • Roles: CEO / Founder, Head of Sales, Product Manager

Generic recommendation engine built on cbruyndoncx/awesome-comparisons.

How you ask

  • “Recommendation engine”
  • “Compare and recommend”
  • “Shortlist”

Inputs

  • vertical_id — kebab-case dataset id (e.g. uae-health-insurance, ai-cli-agents)
  • source_corpus — folder of source spec files (PDFs) to convert with docling
  • vertical_id — existing dataset id
  • changed_specs — list of source filenames that changed
  • vertical_id — dataset to query
  • profile — hard_criteria + optional weights overrides + recipient block (client_id or lead_id)
  • top_n — shortlist size, default 3

Outputs

  • ..//awesome-comparisons/datasets/{vertical_id}/
  • complete dataset folder ready for awesome-comparisons build
  • dataset folder with only the changed entries rewritten
  • see Output Locations)
  • see Output Locations)` — PDF deliverable or proposal in the shortlist_to_pdfreport.py layout — summary, profile, weights, ranking, comparison grids, per-option details, recommendation
  • see Output Locations)` — ranked shortlist with score, why-it-fits, trade-off