Should companies buy voice bots or build them? Let’s dive into the core components of an effective voice bot, what should be configured (and by who), and how you should think about integrating bots into the customer process. Also, let’s explore significant industry changes and how they may affect voice bot decisions.

Here’s a recap of the key ideas presented during the panel at the VOICE Global conference in June:

• Conversational design is still a new domain, and there aren’t many experts. Also, it’s easy to fake being an expert.
• Knowing what to build is as important as knowing how to build it.
• Contextual information about conversations is critical when designing a customer experience.
• A best-in-class approach to all the different cognitive services you need can be helpful (e.g. use the best speech recognition, the best text-to-speech services, the best NLU capabilities, etc.) and are a fast track to a successful deployment.
• In reality, the build vs. buy dilemma is an outdated question. A more optimized approach is what’s needed. Quickly assemble and orchestrate multiple services together while protecting the ability to iterate on the user experience.
• A common mistake when building conversational experiences is presuming to know how users will interact with your system. Those insights don’t come until you start collecting data. The faster you get users interacting with designs and quickly iterating on those experiences, the more chances you’ll have to be successful.
• Black box solutions stifle your innovation roadmap by tying your fate to another company’s development timeline. Outsourcing bots this way is risky because deployment tends to be mission-critical. You need control over your own development cycles.
• Open platforms are a more competitive option. Focusing on API-native cloud-first solutions that let you integrate best-in-class tools and iterate quickly are the true accelerators toward digital transformation. Build bots that work cooperatively across an open ecosystem, not alone in silos.