Customer support automation can turn your customer service into a competitive advantage by meeting your customer where they are, and providing fast, personalized service 24/7. But is it better to go the generative AI route (and get up and running in minutes) or create individual dialogue flows from scratch (that increase personalization and overall automation rates)? Or is a combo of both the best thing for your business?
In this guide, we’ll go over these two primary approaches to automation — generative AI vs. conversation design — and help you understand when to use each, and determine the approach to bot building that’s best for your business.
Generative AI bots
Generative AI-powered bots are the coolest new kids on the block and for good reason: ChatGPT showed the world what chatbots were really capable of and turned the way we think about automation on its head. Generative AI bots are powerful and instantly effective. They are able to synthesize information very quickly and converse in a natural-sounding, human-like way.
When to opt for generative AI-powered support automation
If you’re just getting started with automation, generative AI is probably the approach for you. You’ll get the fastest time to value since they can be up and running in minutes, so you’ll immediately start deflecting questions away from your agents, reducing backlog, and more.
Secondly, if you are simply looking to use a bot as a first line of defense that can deflect FAQs, a generative AI chatbot is the way to go. Where bots that use conversation design have more flexibility to personalize conversations and automate complex processes, gen AI bots are great at answering the simple questions your agents just don’t have time for.
Whether you’ve already started automating with a traditional conversation design bot or not, another sign that gen AI could benefit your organization is if you have a robust knowledge base. Because gen AI bots work by analyzing existing data and information and then summarizing and supplying the answer, they tend to work well for businesses that have all that data documented already. If your help center is up-to-date, doesn’t contain conflicting information, and answers all of your most frequently asked questions, your bot will be set up for success from the get go.
Meet Ultimate’s generative AI offering: UltimateGPT
Our generative AI bot works by analyzing an existing knowledge source — like your Zendesk help center — to instantly and accurately answer your customers’ questions. You simply connect your public help center and UltimateGPT will generate a custom chatbot for you in minutes. Try it out for free here.
More ways to leverage ChatGPT for customer support
Conversation design-based bots
Conversation design bots are the tried and tested automation tool of choice for leading companies. They differ from generative AI bots in that a human has to create dialogue flows to lead customers through resolutions with your bot. So instead of your bot being able to work instantly, it usually takes a couple weeks to sort out the most frequent intents (reasons your customers contact you) and create conversations for each of them. However, this means they’re able to deliver more personalized interactions, and furthermore, you can automate much more complicated resolutions.
When to choose conversation design and intents-based automation
Conversation design-based bots are the right choice for you if you have lots of historical data to build on, but your knowledge base isn’t up-to-date. Historical data means AI can analyze what your customers are asking most frequently. This gives you a data-driven roadmap of which intents to automate first.
Use our free CS Automation Explorer tool to learn your support automation potential.
Conversation design bots will also work best for you if you’d like complete control over the conversations your bot has with your customers. Because a human will be writing the dialogues and designing the flow of conversation, you can ensure the bot responds exactly as you’d like it to.
This type of bot is also best suited for those that want to enable more self-service and automate more complicated requests. You can integrate your back office systems to your bot. This means they have access to all the same information your human agents do, so they can fetch data from the rest of your tech stack to personalize conversations, update information, change and cancel bookings and much more.
This intents-based approach is also great if you’ve already started with generative AI and you’ve seen what it can do, but you’d like to increase your automation rates. You can add on intents to an existing generative AI bot to build out flows that your current chatbot always gets stuck on.
You can even use both gen AI and dialogues within the same conversation to reduce work for your bot builder. If the answer is in your knowledge base or the customer’s message is simply small talk, your bot builder can opt to have generative AI answer instead of writing out an answer from scratch.
Check out more use cases for generative AI in our Humanized AI series.
The best of both worlds: Hybrid generative AI and conversation design bots
Savvy readers are probably already asking the question: why not both? And you’re exactly right. By combining both technologies, you’ll see the fastest time to value, reduce the total cost of ownership, and increase your automation rate all around.
Find out more about the hybrid approach to automation.
Meet Ultimate’s holistic support automation platform
Our customer support automation platform offers the combination of generative AI and conversation design with every bot. You can choose to get started either way: by building out intents or by having generative AI generate answers to your FAQs and then building from there — across every channel you offer support on. It all depends on which approach you think works best for your business.
Which approach is best: Generative AI first or conversation design first?
We’ve gone over the pros and cons of each approach, so which should you start with? It all depends on where you are in your automation journey, what kind of existing data you have, and how much control you want over the process.
- If you’re just starting out with automation: try generative AI first to instantly generate a custom AI chatbot that answers your customer’s FAQs.
- If you have a robust help center that is up-to-date and answers your most frequently asked questions: start with generative AI.
- If you have an out-of-date help center but have a wealth of historical support conversations: start with conversation design to build out intents that correspond to your FAQs.
- If you want to strictly control exactly what your bot says and does instead of letting AI control the flow of conversation: start with conversation design.
Ultimately, of course, the most efficient and effective bots will use a combination of both technologies, but it’s all about deciding what works best for your team.