Conversational IoT and AI powered by “chatbots”
Even in industry, branding and personality play a role
On January 27, CFE Media and Plant Engineering presented a webcast on the topic of “Conversational IoT and AI for industrial environments.” The presenter was chief growth officer of UIB, Ken Herron.
Conversational IoT uses a natural-language processing AI engine to allow human-like interactions through voice and text on any communication channel with any connected device, system, or software with an available application programming interface (API).
The presentation looked at common use cases for conversational IoT, including for safety, secure access, operator training, predictive maintenance, consumables, machine synching and production reporting.
Following is a selection of questions Herron answered as part of the webcast.
Question: Is it (conversational IoT and AI) right for us?
Herron: It depends. Ask yourself, “What is the specific business problem I am trying to solve? Is it a problem that a conversational IoT or conversational AI solution can solve?” Pro Tip: Ask yourself if there’s an approved budget to fix this particular business problem!
Question: How long will it take?
Herron: It depends. The size and scope of your conversational IoT or conversational AI project will determine how long it will take. The primary determiners of how long it will take are the number (and which!) integrations and the number of intents your project needs.
Question: What’s the return on investment (ROI)? What’s the up-front investment?
Herron: It depends. The size and scope of your conversational IoT or conversational AI project will determine both the up-front investment and the ROI. Pro tip: It’s cheaper to host on a public cloud (i.e., AWS) than it is to host on-premise.
Question: Where should we start?
Herron: Start with answering what you want to achieve so that your conversational IoT or conversational AI project solves that specific business problem. Pro tip: Get internal alignment/consensus from all stakeholders (i.e., IT, finance, sales, HR, and any other impacted organizations) up-front.
Question: What if our users switch to different channels?
Herron: They will. Plan on your users switching to different channels. Future-proof your conversational IoT or conversational AI solution. How? Make it omnichannel from the start, allowing you to add new channels in real-time.
Question: Can you please explain how a chatbot can be branded and have a personality?
Herron: Branding and personality drive a chatbot’s success. Great chatbots speak in the brand’s voice and tone and have a personality aligned with the chatbot’s users to maximize engagement and use.
Question: What if a channel is banned/goes down?
Herron: They will. Plan on communications channels being down and banned. Ensure business continuity for your conversational IoT or conversational AI solution by making it omnichannel from the from the start, allowing you to add new channels in real-time.
Pro tip #1: Encourage your users to speak with your chatbot on multiple communications channels on day 1, so there’s always at least one backup channel available.
Pro tip #2: Don’t forget that WhatsApp, Facebook Messenger, and Instagram have gone down simultaneously, so be sure to include communications channels outside the Meta universe.
Question: How many variants should each question have?
Herron: It depends. Each question is different. Some questions need 2-3 variants; other questions need 20 variants. Pro tip: Use a two-week testing period of your conversational IoT or conversational AI solution to let the users define the variants. Pay attention to how users ask questions — those should be your variants!
Is the communication protocols or language created to allow for human and machine interface typically created by onsite company personnel or is it best created by software companies and developers?
Almost every company asks this make vs. buy question, and the answer almost always comes down to technical resources. Are your expensive technical resources’ time best spent working on your conversational IoT or conversational AI project or their “day jobs.”
Pro tip: If the answer to this question is “no,” don’t create your solution in-house — “Do you have at least one person on your team with 2+ years of experience working with cloud-based REST APIs, NLP AI engines, and chatbots that can be fully dedicated to the project?”
Question: Should our conversational IoT or conversational AI project use voice or text?
Herron: The best practice is for *every* conversational IoT and conversational AI project to use both voice AND text.
What are the most common mistakes companies make when creating chatbots?
- They develop them without understanding their users’ needs.
- They have no key performance indicators (KPIs), no tracking and no ROI.
- It’s not on the users’ preferred channels and in the users’ preferred languages.
- It’s treated as a “set it and forget it” project vs. the opportunity to actively listen at scale.
Question: What’s your favorite pro tip for people creating their first chatbot?
Herron: Use just half of your project’s budgeted intents at launch. Regardless of the use case or industry vertical, every chatbot owner wants new/additional intents after they launch their chatbot.
Original content can be found at Plant Engineering.