Procurement with AI
Cimple , Cimple , Cimple , Cimple , Cimple , Cimple , Cimple , Cimple , Cimple ,
Cimple , Cimple , Cimple , Cimple , Cimple , Cimple , Cimple , Cimple ,
Introduction
Cimple, a growing ProcureTech marketplace, helps governments and enterprises leverage AI to streamline procurement and unlock greater supplier value. A common challenge for their enterprise customers is poor catalogue data quality, with incomplete pricing, product details and availability - leading to inefficiencies, missed savings opportunities and operational waste.
To address these issues, Cimple partnered with Studio Graphene to develop a proof-of-concept (POC) solution aimed at improving the quality of supplier catalogue data. This collaboration focused on overcoming the challenges of incomplete or inaccurate product information, a major roadblock for procurement teams, especially within the NHS.
.avif)
The solution also addressed the absence of product images, a key factor in influencing buyer decisions.
Leveraging advanced prompt engineering and OpenAI models, the Catalogue Fixer enabled the scalable enrichment of product data, ultimately driving efficiency in the procurement process.
Studio Graphene used purpose-built AI tools to create the "Catalogue Fixer," an engine designed to collate, cleanse and augment supplier data from various sources.
The solution also addressed the absence of product images, a key factor in influencing buyer decisions. Leveraging advanced prompt engineering and OpenAI models, the Catalogue Fixer enabled the scalable enrichment of product data, ultimately driving efficiency in the procurement process.
.avif)
.avif)
Results
Conducted manual interactions with LLM chatbots (ChatGPT, Gemini, Perplexity) to identify the most effective models for data enrichment and retrieval.
Developed a backend engine leveraging Python, Flask and OpenAI APIs to process and standardise supplier data at scale.
Achieved a 60% success rate in enhancing product data accuracy on the first attempt through advanced AI techniques, with defined next steps to further improve performance.
Delivered a scalable solution to overcome fragmented and inconsistent reference URLs, ensuring data consistency.
.avif)
