The Twilio Autopilot Sunset is Upon Us
Later this month, as of February 25, 2023, Twilio Autopilot will be at end-of-support, with complete end-of-life scheduled for August of this year. Autopilot was announced in 2018 and originally went GA in 2019, and provided Twilio customers with developer-friendly tools for integrating Conversational AI into chatbots and automated voice applications.
VPS was able to integrate Autopilot into a handful of customer solutions over the past few years, so we definitely saw the potential. Compared to rigid syntax-matching rules that are required when building an IVR or bot flow with Twilio Studio, for example, Autopilot provided tactical flexibility for understanding more open-ended requests from end users. For example, we saw success using Autopilot when building an IVR that needed to capture specific place names. Capturing proper names is difficult with standard speech recognition that is bundled directly with Twilio Voice.
Autopilot also stood out when faced with use cases where it made sense to give an end user the option to state their intent, and not be forced to choose from a set menu. This isn’t always the best design pattern, but does make sense when there are more than 5 or 6 different options or paths. We could build different matching rules with synonyms, and Autopilot’s AI was very accurate when determining possible variations.
Unfortunately, I think a couple things led to Autopilot’s downfall. First, I think it comes down to a matter of focus for Twilio. Twilio’s real focus is providing reliable and flexible communication capabilities with the cloud. Programmable Voice, Programmable Messaging, Flex for the Contact Center, and below it all – The Twilio Super Network. It’s not surprising that maintaining the infrastructure required for providing a developer-friendly AI platform was a bit out of the norm for the team behind the scenes for Twilio. Second, the interface for configuring and managing autopilot was a bit different than the rest of the Twilio Suite. It required a unique UI for managing tasks and training in the Twilio Console which made it less familiar to developers while also being a bit too complex for a novice administrator. The API-based interface also followed different patterns compared to the rest of the Twilio API, which made it confusing for anyone trying to work with it from a purely code-based direction.
Although the departure of Twilio’s Autopilot product may come as surprising news, VPS is still committed to providing reliable support services and ongoing guidance to all existing Autopilot customers. Twilio’s official recommendations for impacted customers are available here, but this information doesn’t provide the full picture for every customer. That’s where VPS comes in.
VPS has compiled a list of our own migration recommendations for different scenarios. We do support the use of Google DialogFlow in many circumstances, including leveraging Twilio’s new <VirtualAgent> Twiml Tag. This does require customers to bring their own Google Cloud account, so billing will need to be set up for both Twilio and Google.
Times like these are when it’s valuable to have a Systems Integrator partner on speed dial. Vision Point Systems works with many leading technology platforms, and it’s our role to advise and guide our customers to whatever solution is best for them. Twilio is still the market leader for programmable communications, and VPS will continue to deliver contact center and customer engagement solutions with Twilio at their core.
If you’re still running Autopilot, now is the time to move to something new. With Twilio dropping support at the end of the month, any support for developers or data managers on systems using Autopilot will have to come from partners like VPS. After August, the plug is pulled, and the systems will just stop working. Hopefully replacement systems can be built in a few weeks or months, but now is the last chance to be safe and not sorry.
As a trusted Twilio service partner, VPS offers customized development solutions to meet the unique needs of each client. Our objective is to leverage automation technology and improve the customer experience, service quality, and cost efficiency.
Author
