From 7ac18a486b96034ecd63884efe4f7f4d75d6bd93 Mon Sep 17 00:00:00 2001 From: Steve Jensen Date: Mon, 2 Oct 2023 08:29:38 -0500 Subject: [PATCH] doc updates --- README.md | 48 ++++++++++++++++++++++++++---------------------- 1 file changed, 26 insertions(+), 22 deletions(-) diff --git a/README.md b/README.md index bd88de6..58a867d 100644 --- a/README.md +++ b/README.md @@ -3,35 +3,39 @@ ## Background This project is a [SIP Media Application](https://docs.aws.amazon.com/chime-sdk/latest/ag/use-sip-apps.html) and makes use of the -[Java Chime SMA Flow Library](https://github.com/docwho2/java-chime-voicesdk-sma) to deliver a [ChatGPT](https://openai.com/chatgpt) voice bot IVR application. The IVR application is integrated with the [Square API](https://developer.squareup.com/us/en) to allow callers to ask questions about products -and business hours, tranfer to employee cell phones, etc. +[Java Chime SMA Flow Library](https://github.com/docwho2/java-chime-voicesdk-sma) to deliver a [ChatGPT](https://openai.com/chatgpt) voice bot IVR application. The IVR application is integrated with the [Square API](https://developer.squareup.com/us/en) to allow callers to ask questions about products +and business hours, transfer to employee cell phones, etc. ## Use Case -[Copper Fox Gifts](https://www.copperfoxgifts.com) is a retail store located in a small town in MN. The goal is to field the majority of the calls without human intervention. -- Over 50% of the calls are "Are you open now?". This is a typical call in resort type towns where hours change seasonally and frequently. Many people always call ahead and want to know if the store is open before they leave. -- The follow on is always, OK so what are the hours then. -- 45% are "Do you have XYZ product?" (mittens, hats, gummy bears, shorts, candles, etc.) -- The remaining calls are from vendors working with someone at the store and want to talk with person XYZ. +[Copper Fox Gifts](https://www.copperfoxgifts.com) is a retail store located in a small town in MN. The goal is to handle the majority of calls without human intervention. Here's a breakdown of the types of calls they receive: + +- **Over 50% of the calls**: "Are you open now?" This is a typical question in resort-type towns where store hours change seasonally and frequently. Many visitors prefer to call ahead to ensure the store is open before they head out. + - A common follow-up to this is: "What are your hours today?" + +- **45% of the calls**: "Do you have XYZ product?" Queries range from mittens, hats, and gummy bears to shorts, candles, and more. + +- **The remaining calls**: These are primarily from vendors who are coordinating with a specific individual at the store and wish to speak with them directly. ## Solution Summary -The goal is to provide a "Store Virtual Assistant" based on [OpenAI ChatGPT](https://openai.com/chatgpt) that not only can answer store specific questions, but anything in general the caller asks. -- Use of [ChatGPT Function Calls](https://platform.openai.com/docs/guides/gpt/function-calling) to enable Sqaure API calls to access inventory, employee, and store hours. - - Further use of function calls so the model can indicate when a call needs to be transferred or ended. -- Strategic prompting to prime the model with store information and how to interact with the caller. -- A solid and reliable solution deployed multi-region within AWS and completely cloud based. - -Features: -- Callers are always greeted and told whether the store is open or closed. -- Store hours are based on real-time data from a Square API call. -- Product category and item searches are based on Square API calls. - - Callers can ask what types or specific products are available. -- Speaking with employees or main store line. - - Caller can ask to speak to specific employees and this information is based on a Square API call (Team Member list) - - If the caller just wants to speak with a person, the model is primed with a general number to transfer the call to. - - When tranferring to the main line, this transfer is optimized to use SIP direct to the store [Asterisk PBX](https://www.asterisk.org). +The goal is to introduce a "Store Virtual Assistant" powered by [OpenAI ChatGPT](https://openai.com/chatgpt) that can not only answer store-specific queries but also address any general questions the caller might have. +- Utilize [ChatGPT Function Calls](https://platform.openai.com/docs/guides/gpt/function-calling) to facilitate Square API calls, enabling access to inventory, employee details, and store hours. + - Further leverage function calls so that the model can determine when a call should be transferred or concluded. +- Employ strategic prompting to prime the model with pertinent store information and to guide interactions with callers. +- Ensure a robust and dependable solution that is deployed across multiple regions within AWS and is entirely cloud-based. + +**Features:** +- Callers are consistently greeted and informed about the store's operational status (open or closed). +- Store hours are ascertained in real-time via a Square API call. +- Product category and individual item searches are also driven by Square API calls. + - Callers can inquire about product categories or specific items in stock. +- Engaging with store employees or the primary store line. + - Callers can request to connect with specific employees, with the information sourced from a Square API call (Team Member list). + - If the caller simply wishes to speak to a representative, the model is preloaded with a default number to redirect the call. + - During transfers to the main line, this process is optimized to use SIP directly connecting to the store's [Asterisk PBX](https://www.asterisk.org). + ## High Level Components