Conversational AI Developer Tools Compared

12 Apr 2023 | By Tushar Goswami

After OpenAI’s huge success, we’re getting a lot of requests from customers to use Generative AI to build a Conversational AI for their business.

But is OpenAI the right developer tool for your needs?

Good_Advice_from_OpenAI 1

1. YES! If your AI must

  • Answer questions like “What should i wear on Halloween eve? I’m a girl who likes Marvels, and I want to be a character that is genius”.
  • This will get your user advice from “collective intelligence of humans” who’ve written billions of webpages on internet. That OpenAI has learnt from.
  • Generative AI developer tools like OpenAI ChatGPT, or any LLM like Google BARD, NVIDIA Nemo are good at advising based on collective intelligence of humans.

2. NO! if your AI must

  • Take requests like “shoes under $100 for my Shuri superhero dress. My size is 10 UK size”
  • Your User expects an answer returned from INSIDE YOUR DATABASE.
  • OpenAI would rather give “advice” based on collective genius of billions of webpages. You don’t want to lead your shopper to see the options that your store doesn’t have. Developer Tools based on Knowledge Graph like DiffBot or TalkingDB are good at this, not OpenAI.
PoorSearchResultsbyOpenAI 2
Flowcharts_cannot_be_followed_by_OpenAI

3. NO! if your AI takes must

    • Take requests like “Please take my order but would it reach me before Halloween”
    • Your user expects followup questions (such as customer address), confirms from API if shipment would reach before time, basically follows a FLOWCHART to onboard customer by asking their personal details, shipping address, adding products into their cart, before finally taking payment.
    • OpenAI can’t follow flowchart that follows your business process. Developer Tools based on Flowchart like Google Dialogflow, IBM Watson Assistant, or Hybrid.Chat are good at this.

Which of the 3 use cases you want to build in your Conversational AI? Our AI Agency helps you build all 3. First we do a detailed Conversation Design – outlining all the various utterances and response that bot and user would exchange. And then we identify which type of request must be diverted to which developer tool

Several businesses are making the wrong choice of picking Generative AI to build “everything” in their chatbot. Lets stop that. Spend time in conversation design – just like you used to spend time in UI Design in the good old days of building GUI applications 🙂 This will help you get clarify on the use cases your conversational AI needs to offer.

And then right developer tool to solve the right use case. For all those who fancy an infographic to compare developer tools, here is one. Please respond back with comments, and help us make this infographic better

Strategy_Canvas_for_Conversational_AI_Developer_Tools

FoodYo: Food ordering bot