Q1 2026 — 6th Consecutive Quarter
Adoption Increase
Generally Available Today
in 2025
You've been there. You explained the whole situation to the chatbot. You got transferred. The agent picked up and asked you to start over. That moment — that specific frustration — is something almost every customer has felt. And at SIGNAL 2026 in San Francisco today, Twilio made a direct bet that fixing it is an infrastructure problem, not a training problem.
Not a better script for agents. Not a smarter chatbot. The fix has to happen underneath all of it, at the layer that every channel sits on top of. That is what Twilio shipped today.
What Actually Shipped — and It's All GA
Worth saying clearly upfront: this is not a roadmap. All four capabilities are generally available as of today, May 6, 2026. Private beta customers including Car Finance 247, Centerfield, and Constellation Dealerships have been running on this since January.
Conversation Memory is the one customers will notice first. It builds a living, identity-resolved profile of each customer, pulling together conversation history, preferences, behavioral signals, and current state. That profile is available to both human agents and AI agents in real time. Every conversation picks up where the last one ended, whether the customer reached out via voice, SMS, WhatsApp, or chat.
Conversation Orchestrator is the routing and handoff engine. It connects interactions into a single thread as they move across channels, and manages the handoffs between human agents and AI agents without dropping context in the process.
Conversation Intelligence watches live interactions in real time, surfacing sentiment, intent, friction signals, and suggested next steps. It uses generative AI language operators to turn raw conversation into actionable intelligence while the call or message thread is still happening.
Agent Connect is the strategic tell. It is a self-hosted, model-agnostic SDK in Python and TypeScript that connects any AI agent, from any vendor, directly to Twilio's voice and messaging channels. OpenAI, Anthropic, Azure, AWS Bedrock, LangChain, in-house models. Bring whatever you're running. Twilio handles the channel access, the session management, and the customer memory.
"Most brands still treat every conversation with a customer like it's the very first one. Twilio is changing that at the infrastructure layer, so every business built on Twilio can remember, learn, and respond like they actually know their customers."
Bring Your Own Agent. We Have the Data.
Agent Connect deserves a closer look because it signals something important about where Twilio is positioning itself.
Twilio is not trying to win the AI model competition. That race has too many well-funded players and the landscape is shifting too fast. What Twilio is saying instead: pick whatever model works for you, we'll be the layer underneath that actually knows your customer. Deploy GPT-4o today, switch to something else next year. You don't have to rewrite your channel integrations every time.
That is a smart bet right now. One of the real blockers to enterprise agentic AI rollouts has been model lock-in anxiety. Teams don't want to build deep on a specific AI vendor when capabilities and pricing are moving this fast. Agent Connect removes that blocker by making the model layer swappable while making the Twilio layer the thing that accumulates value over time.
Twilio lowers the switching cost on the AI side. It raises it on the data and channel side. That is exactly the position they want to occupy.
Why This Move Makes Sense Now
I covered the Q1 2026 earnings story here last week. Six consecutive quarters of 20% voice revenue growth. Net revenue retention at 114%. Multi-product adoption up 29%. CEO Khozema Shipchandler told investors on that call that the SIGNAL announcements would be the most consequential product moment in the company's history.
The most telling line from that earnings call was this one: "Customers no longer view Twilio as just a provider of communications channels." The SIGNAL launch is the product that earns that sentence.
Here's the business model reality underneath all of this. A pure communications platform business — deliver SMS, make calls, send email — is increasingly hard to differentiate. Multiple vendors do it reliably at scale. Pricing pressure is real. The end state of that competition is commoditization. Twilio's pivot to persistent customer memory and conversation orchestration is a direct response to that pressure.
Once the memory layer holds the institutional knowledge of every customer relationship you have ever had, the conversation about switching vendors changes completely. It is no longer a question of porting API calls. It is a question of what you lose in customer context when you move.
The Part That Needs to Play Out
The architecture is solid. The GA delivery is real. The risk I'd watch is adoption pace in existing enterprise environments.
To get the full value of Conversation Memory, you need to route your interactions through Twilio's infrastructure. That is manageable for companies already deeply on Twilio. For enterprises with entrenched Genesys or NICE contact center deployments, or years of homegrown routing logic, the integration work is real. The good news: Conversation Orchestrator is designed to work alongside existing deployments without requiring a full re-architecture. And Agent Connect's incremental model means you can start with one use case and build from there rather than betting the whole stack on day one.
NICE, Genesys, and Five9 are building toward the same destination, but from the agent desktop inward. Twilio starts from the channel layer, the place where every interaction actually transits. That is a real structural advantage. Whether it becomes a durable one depends on how fast enterprise CX teams actually adopt the memory layer.
If Twilio accumulates persistent memory of every customer conversation across every channel, what does your next contract negotiation look like — and have you modeled the switching cost you are building into your stack before you need to use it?
The shift from channel vendor to memory infrastructure changes who holds leverage at renewal. Evaluate Agent Connect's model-agnostic promise in your specific deployment — the openness is real, but the memory accumulation is the lock-in.
Works Cited
Twilio. "Infrastructure for the Agentic Era: Everything We Launched at SIGNAL 2026." twilio.com. 6 May 2026.
Twilio. "Introducing Conversation Orchestrator." twilio.com. 6 May 2026.
Twilio. "What Is Twilio Agent Connect?" twilio.com. 6 May 2026.
Twilio. "Real-Time Conversation Intelligence Is Now GA." twilio.com. 6 May 2026.
Business Wire. "Twilio's Next Generation Platform: An Infrastructure Layer for Every Conversation in the Agentic Era." businesswire.com. 6 May 2026.
CMSWire. "Twilio Bets Its Infrastructure Layer Can Own the Agentic Customer Conversation." cmswire.com. 6 May 2026.
Shipchandler, Khozema. Q1 2026 Earnings Call. Twilio, Inc. April 2026.
