To get the best results from OI-Q, try to be as specific as possible with your prompts. The clearer your instructions, the more accurate and helpful the AI will be!
Understanding OI-Q
OI-Q is a smart AI tool that helps you uncover what is happening in customer transcripts. It works within the transcript you select and doesn’t pull in outside info.
To get the best results, follow these tips:
Be clear & specific – Ask direct questions as the more precise your question, the better the answer.
“Did the customer appear frustrated on this call? Tell me on 50 words or less.”Provide context – If background details help, include them in your prompt.
“This customer experienced a failure with our online app, identify what we could do to improve the app”Define your format – Need bullet points or a specific structure? Let OI-Q know.
“How could this call could have been handled better. Format your response with 3 bullet points of top opportunities”
Extract verbatim information – If you need something quoted word-for-word, say so.
“Tell me the exact words the customer used when they described their product issue”Use examples – Provide reference terms (e.g., "Example product names: XYZ").
“Does this customer mention xyz product?”
Using OI-Q² for Summarization & Multi-Select Queries
OI-Q² is designed to summarize multiple transcripts into a single response, helping identify common themes.
Keep these tips in mind:
Ask one question at a time – Since OI-Q² processes the full conversation history, running it multiple times may impact accuracy and cause errors.
Keep it short – Adding "summarize in 50 words or less" into your prompt helps keep responses focused.
Be mindful of limits – OI-Q has a ~32,000-token limit (input + output). So if you input 100 calls with a limit of no more than 200 words in your prompt, then the input is ~20000 tokens.
“Analyze these 50 interactions, and tell me the frequency of which xyz occurs. Respond in 50 words or less”
Advanced Uses & Cool Features
Estimate time spent on tasks – OI-Q can calculate how long a specific topic takes in a call.
“Tell me how long it took the agent to verify the customer on this call”
Check calculations – If you need percentages or math-based answers, ask OI-Q to verify them.
“Analyze these 75 interactions, and tell me in a percentage how often customers mentioned xyz. Respond in 50 words or less”
Creative formatting – Want your answer as a poem, song, or script? Just ask!
“Summarize this interaction as a Haiku”Big-picture insights – Coming soon, OI-Q will handle thousands of inputs to create knowledge articles, FAQs, and more.
What OI-Q can’t do (yet!)
Remember that OI-Q does not have access to metadata, and therefore can’t retrieve agent names, handle times, or other metadata (for now).
By following these best practices, you’ll get the most reliable and useful insights from OI-Q and OI-Q².
Happy prompting! 🚀