I attended Twilio SIGNAL Tokyo 2026 held on October 8th
This page has been translated by machine translation. View original
Hello, I'm Subaru.
I attended Twilio SIGNAL Tokyo 2026, held on October 8th! This time, I'd like to go over the overview and content of this event.
About the Event
Twilio SIGNAL Tokyo 2026 is the Japan edition of Twilio's annual conference "SIGNAL," marking its first-ever Japan event. With the theme of customer engagement in the AI era, the event featured a keynote speech by Twilio's CEO, panel discussions by companies utilizing Twilio, product sessions, and more.
- Time 16:30～18:30
- Venue Grand Hyatt Tokyo
About the Sessions
CEO Keynote: The Future of Customer Engagement
Twilio's CEO will unveil "Agentic AI (Autonomous AI)," the most important theme of 2026, along with the latest roadmap.
This session was delivered by Twilio's CEO, Khozema Shipchandler.
He noted that while AI is increasingly being utilized in various settings, the standards for customer experience have risen accordingly. In my own work, I sometimes hear about challenges related to contact centers, and I've felt that expectations for AI-powered customer experiences are higher than they used to be. It's also predicted that by 2029, approximately 1 billion AI agents will be deployed across many companies, with most of them interacting with people. In this context, Twilio is positioned to support the next generation of customer experience, and this year marks the announcement of the next-generation Twilio platform as well.
Special Keynote: Digital Transformation on a National Scale
We welcome Takuya Hirai, Japan's first Minister for Digital Transformation, to discuss the importance of breaking free from legacy systems, building government services that are accessible to everyone in an aging society, and achieving a proactive, safe, and seamless "Society 5.0" through cloud communications, AI, and the My Number Card.
This session was delivered by Takuya Hirai, Japan's first Minister for Digital Transformation.
He noted that while the internet spread worldwide because its convenience far outweighed its risks, with AI, one misstep could bring its convenience and risks to a balance. That's precisely why it's important to clearly define "what needs to be protected," "what kind of world we're aiming for," and "what means we choose to get there" — and he mentioned that Twilio has a clear vision of the world it aims to create. He also said that our relationship with AI should not be "human-in-the-loop" but rather "human-in-the-lead." I interpreted this as not just having humans serve as checkers within AI processes, but rather humans holding the reins — like a dog's leash — deciding what to entrust to AI and who bears responsibility.
Cybersecurity is also an extremely challenging issue at the national level, and going forward, it will be necessary to choose trustworthy options. He also spoke about how infrastructure like Twilio will be needed to build a system that people can use with peace of mind, while preserving the openness that makes the internet great. Finally, he introduced the idea of firmly protecting what must be protected — such as citizens' rights and privacy — while flexibly reviewing other rules through AI adoption. He concluded by saying that while the risks are significant, the opportunities are many times greater. I felt that AI adoption ultimately comes down to trial and error, and that the knack for it comes through actually using it.
UX is something I'm particularly conscious of in my work, and from a UX perspective, I think it becomes clearer through use where "humans are better," "AI is better," or "both are good."
CX Infrastructure Supporting Growth: The Modernization of Customer Experience at Japanese Companies
This was a panel discussion featuring All Nippon Airways Co., Ltd., Excite Co., Ltd., and ITANDI Co., Ltd. Although each company is in a different industry, I found it interesting how each had challenges unique to their industry, and how Twilio's high flexibility fit neatly into each situation.
All Nippon Airways Co., Ltd. operates approximately 850 flights per day, with more than 100,000 customers on normal days and over 150,000 at peak times. They currently use Twilio to send SMS and email messages, including booking confirmation notifications and online check-in guidance 24 hours before departure, sending multiple notifications to each customer. Before implementation, the challenge was that they couldn't reliably deliver information during irregular flight operations such as cancellations and delays, resulting in customers arriving at the airport without knowing the situation. Additionally, since each system — reservations, operations, marketing — was handling its own communications, it used to take months to years to start something new.
Excite Co., Ltd. uses Twilio as the calling infrastructure to connect fortune tellers and users in their telephone fortune-telling service. Since building a phone system from scratch is difficult for application developers, using an API was a prerequisite. Their previous system required an initial investment and needed additional equipment each time the number of users grew, but by switching to Twilio's pay-as-you-go model, they reduced annual costs to approximately 30% of what they were before, resulting in roughly 180 million yen in cost savings over the two years following the switch.
ITANDI Co., Ltd. has been using Twilio since around 2014 when it first entered Japan. The real estate industry has a practice called "bukken kakunin" (property confirmation) where real estate companies call each other to confirm whether a property is still available, and they built a system to automate this as their first product. At peak times, the system processed approximately 500,000 property confirmation calls per month, reducing the one-to-one-and-a-half-minute confirmation task per inquiry. It also reduced the burden on real estate staff who were reluctant to handle phone calls.
As final advice to the audience, All Nippon Airways Co., Ltd. emphasized the importance of incorporating global best practices, leveraging trustworthy services, and starting small with what's feasible. Excite Co., Ltd. noted that Twilio has highly extensible APIs and comprehensive documentation, so development can begin immediately with no prior knowledge — when rebuilding from their old platform, a single engineer was able to create a prototype in about two months. ITANDI Co., Ltd. shared that internally they talk about "eliminating the UI." Rather than forcing users to download an app and learn how to use it, they aim for a model where people can accomplish what they want through familiar channels like phone and email, leaving the backend to AI.
I strongly resonated with the comment that the worst thing in UX is being forced to do something. Since having to learn how to use something already places a burden on users, I think it's best when things can be completed entirely through familiar channels like phone or email.
Voice AI and Telephony DX: Realizing Next-Generation Customer Engagement
This was a panel discussion featuring AI Shift Inc., Mobilus Inc., and IVRy Inc. All three companies offer voice AI as their own product. AI Shift Inc. mentioned that while the shortage of contact center staff has been discussed for a long time, the replacement with AI has not progressed easily. Japan is a super-aging society, and since Japan is the only country that uses Japanese, localizing for the Japanese market — rather than simply bringing in overseas AI — is important.
Regarding AI adoption timelines, Mobilus Inc. aims to handle 80% of call volume within about four years, IVRy Inc. expects 50% to 70% within ten years, and AI Shift Inc. aims to achieve around 40% of the total within three years. Mobilus Inc. and IVRy Inc. believe that a human element will remain, while AI Shift Inc. sees a hybrid approach in the near term but is aiming for a future where AI handles everything through voice alone.
Regarding Twilio, Mobilus Inc. entrusts all the communications infrastructure — including phone lines — to Twilio, allowing them to focus their own development on the application side, including AI and management screens. Since voice communications must never go down, they greatly appreciate the fact that it operates stably and scales reliably as usage increases. IVRy Inc. also noted that by entrusting the phone infrastructure to Twilio, they could focus on app and AI development, allowing them to grow to a cumulative total of approximately 70,000 accounts.
I often experience firsthand the difficulties unique to Japanese. Depending on the product, Japanese speech transcription — and, somewhat tangentially, sentiment analysis — can still be challenging in some areas.
Product Session: Realizing Voice AI with Twilio Conversation Relay
Twilio's engineers build a solution combining the latest AI features on the spot in an overwhelming live demo. Witness firsthand the "speed of development" where complex workflows take shape in minutes, and the moment of "tangible innovation" you can start applying at your own company tomorrow.
The demo showcased an AI responding to a scenario where a water leak occurred in the middle of the night, combining phone and LINE to handle the situation. This demo was built using Twilio Conversations and ConversationRelay. Twilio Conversations is composed of four functions: Conversation Orchestrator, Conversation Memory, Conversation Intelligence, and Agent Connect. Conversation Orchestrator handles interactions across channels like phone and LINE as a single conversation. Conversation Memory retains customer information and conversation state, while Conversation Intelligence analyzes conversation content in real time. Agent Connect is the mechanism for connecting your own AI agent to Twilio.
The main feature of this session was ConversationRelay. ConversationRelay is a feature that handles everything from transcribing call audio, generating responses with an LLM, and reading out the generated responses. Since you don't need to build the audio processing yourself, you can focus your development on the interaction with AI.
Next came a case study from Ululur Inc. They provide the phone answering service "fondesk" and the cloud-based IVR "fondesk IVR," having grown from four part-time members to over 7,500 companies today. Even without members who had specialized knowledge of phone lines, using Twilio as their foundation allowed them to develop without an initial investment. This spring, they added an "AI Operator" to fondesk IVR that handles AI responses, built by a single person in about two weeks using ConversationRelay. Building speech transcription, LLM interactions, and speech synthesis from scratch would apparently require considerable cost and time. Going forward, they plan to use Conversation Intelligence and Conversation Memory to deliver responses that also take past call history into account.
I felt that with ConversationRelay, the possibilities for voice AI — and what I'd want to try — could expand even further.
Summary
This time, I looked at Twilio SIGNAL Tokyo 2026, which I attended on October 8th. I'd like to continue participating in future Twilio events and keeping up with updates on an ongoing basis.