Instruction Genius: Customer issue resolution using AI

Solution overview

• Problem

  Legacy-format technical documentation, such as manuals and instructions for use, is often not interpreted or found efficiently by users. Understanding technical device information is exceptionally critical in healthcare settings. Customer inquiries are usually sent to the distributor and/or manufacturer of such devices, where front-line customer service staff triage, consult, and reference publicly available technical documentation.

• Computer vision

  Using multimodal AI, we developed a platform using computer vision with an AI model designed to ingest and transform legacy-formatted technical documentation into a searchable digital format.

• Retrieval-augmented generation

  We answered customer inquiries using retrieval-augmented generation (RAG) with a large-language model (LLM). Based on real data from alpha users, we developed internal benchmarks and concluded that our solution could accurately deflect a significant number of relevant technical inquiries.

• Outcome

  Instruction Genius users are successfully using AI to deflect technical customer service inquiries. Issue deflection with AI results in reduced costs and a better customer experience. It enables humans to be more productive and frees up customer service staff to focus on solving more pressing issues that require human ingenuity.