Generating step-by-step procedures from historical conversation records using a language model

A non-transitory computer-readable medium storing executable instructions, which when executed by a processing device, cause the processing device to perform operations comprising:

• programmatically generate a resolution assessment input instruction for a first language model to cause the first language model to determine resolved historical conversation records, the resolution assessment input instruction generated based at least on historical conversation records and a resolution assessment prompt instructing the first language model to determine whether an issue was resolved in the historical conversation record;
• programmatically generate a procedure generation input instruction for a second language model to cause the second language model to generate procedure documents, the procedure generation input instruction generated based at least on the resolved historical conversation records and a procedure generation prompt instructing the second language model to (1) extract one or more issues from each resolved historical conversation record, (2) generate a corresponding issue description of each issue, (3) generate a corresponding step-by-step procedure resolving each issue, and (4) generate a corresponding issue resolution description for each issue;
• and
• storing the procedure documents in a knowledge base;
• and
• responsive to a query, (1) performing a semantic search to determine a similar procedure document above a threshold semantic similarity to the query based on computing a semantic similarity between procedure embeddings of the procedure documents to a query embedding of the query and (2) causing a representation of the similar procedure document to be presented via a user interface (UI) of a computing device.

A computing system comprising:

• a processor;
• and
• a non-transitory computer-readable medium having stored thereon instructions that when executed by the processor, cause the processor to perform operations including:
• programmatically generate a resolution assessment input instruction for a first language model to cause the first language model to determine resolved historical conversation records, the resolution assessment input instruction generated based at least on historical conversation records and a resolution assessment prompt instructing the first language model to determine whether an issue was resolved in the historical conversation record;
• programmatically generate a procedure generation input instruction for a second language model to cause the second language model to generate procedure documents, the procedure generation input instruction generated based at least on the resolved historical conversation records and a procedure generation prompt instructing the second language model to (1) extract one or more issues from each resolved historical conversation record, (2) generate a corresponding issue description of each issue, (3) generate a corresponding step-by-step procedure resolving each issue, and (4) generate a corresponding issue resolution description for each issue;
• storing the procedure documents in a knowledge base;
• and
• responsive to an ongoing conversation, (1) performing a semantic search to determine a similar procedure document above a threshold semantic similarity to the ongoing conversation based on computing a semantic similarity between procedure embeddings of the procedure documents to a representation embedding of a conversation representation of the ongoing conversation and (2) causing a representation of the similar procedure document to be presented via a user interface (UI) of a computing device.