The shopping carts won the page.
The conversation is still up for grabs.
Web commerce was won by whoever owned the checkout page. The next channel is being won right now, in the conversation, and the incumbents are not in the race.
Executive summary
Web commerce is settled, and the shopping carts won the checkout page years ago.
Buying has moved off the page and into the conversation: people call a number and an AI voice agent answers, or open a chat and an agent runs the thread to the sale. A new channel is never won by the last channel’s stack. It is won by whoever solves the hardest part of it, and in the conversation the hardest part is the payment.
In production with
The page is won. Let it go.
Part 1: The problem
A shopping cart platform sells one thing underneath all the features: the place a merchant runs their storefront and takes a payment on the web. The strongest ones built their own payment product into the checkout, because owning the payment is how a platform stops being a tool and becomes infrastructure.
That worked. It is also finished. The web checkout is mature, the incumbents hold it, and a merchant will not migrate their storefront to capture a few points of payment economics. If the plan is to be a better checkout page than the one a merchant already uses, that is a war that ended before you arrived.
Average documented cart abandonment, across 50 studies from 2006 to 2025. The web checkout, the thing the incumbents perfected, still loses roughly seven of every ten carts.
Baymard Institute, 2025 Buying moved into the conversation
The shift
A customer phones to place an order, change a booking, settle an invoice, or renew a policy. A few years ago a person answered. Today an AI voice agent does, and it holds the whole conversation: the questions, the options, the upsell, the confirmation.
A customer opens a chat, messages on WhatsApp, or replies to a marketing thread. An AI chat agent answers, recommends, handles the objection, and brings them to the point of sale, all inside the thread, without a tab ever opening.
These are not support tickets. They are sales, often the higher-intent ones a customer wanted to talk through before committing. The constraint that kept these channels small, a trained human on every interaction, is the constraint that just lifted. The agent handled all of it, right up to the moment money changes hands.
A sales channel with no checkout
The wall
The agent that just closed the sale cannot take the payment. So one of three things happens, and all three leak. Every one is a sale that was won and then degraded at the last step.
Transfer to a human
Throws away most of the economics that made the agent worth deploying.
Read the card aloud
Drops card data into a recording or transcript and creates a compliance problem the business may not know it has.
Push back to a web page
The exact friction the conversation removed. Against a 70% abandonment rate, a measurable share never complete.
This is the shape of problem the web solved twenty years ago. A storefront that could show the product but not take the money was not a store. The conversation is at that exact stage now: it can do everything except finish. The missing piece is the close.
The numbers, from independent sources
Part 2: The evidence
of customer service journeys will begin and resolve inside conversational assistants by 2028.
Gartner, 2025of common service issues resolved autonomously by agentic AI by 2029, cutting service cost ~30%.
Gartner, 2025of phone leads convert during the call, across 60M+ analysed calls.
Invoca, 2025voice commerce market by 2030, up from $42.75B in 2023 (24.6% CAGR).
Grand View Research, 2024conversational (chat) commerce market in 2024, growing ~15.6% a year.
Market.us, 2025conversion lift reported by insurers using AI voice and chat assistants.
McKinsey, 2024It is already deployment, not forecast: Yum Brands rolled AI agents to 500 Taco Bell and Pizza Hut locations in 2025, and Wendy’s expanded FreshAI toward 500 to 600 locations. Businesses expect to add about four new channels, with 95% maintaining or increasing their channel count (Twilio, 2024), and 73% of shoppers already use more than one channel per journey (Zendesk, 2024). Every channel is another surface where a sale can close, and another that needs a way to take the money.
Part 3: The hypothesis
A new channel is not won by whoever shows up with the old channel’s tools.
It is won by whoever solves the part everyone else finds too hard. Agents are on their way to being a commodity, so competing on the agent is competing on a feature everyone will soon have. The hard part is the payment, and that position is currently empty.
Why the payment is the hard part
Three reasons that do not exist on a page
The card cannot be exposed
On a voice call the number cannot land in the recording or the model’s context. On chat it cannot touch the transcript. This is PCI with teeth, in exactly the channels that are opening up, and it is unforgiving.
It has to settle on the merchant’s own rail
The merchant already has a provider, a negotiated rate, a settlement account, sometimes one their compliance team mandated. The voice and chat sale must land where the web sale lands. A channel that forces a new provider is a migration, and merchants do not buy migrations.
It has to work across every surface
Voice, chat, SMS, and the handoffs between them each capture payment differently. A customer should be able to start on chat, move to voice, and have the cart and the card token follow.
Why the obvious answers fail
Each for a specific reason
Lives on the page, and is the platform’s own provider by design. It fails the merchant’s-own-rail test before the channel question even arises.
One channel on one provider takes a quarter. Then the second provider, the second PCI surface, the Level 1 obligation, the audits, and a payments team nobody set out to hire.
The shortcut that loses the enterprise accounts, because their provider is not negotiable.
Closes one sale in one channel, carries no identity across a handoff, and pushes reconciliation back onto the merchant. A patch, not a position.
Where payments actually sit
Part 4: The solution
A payment layer sits between software and the payment providers: thin where a PayFac is heavy, provider-neutral where a gateway is single-provider, channel-native where an orchestrator is page-bound. It exposes the merchant’s existing payment relationship to wherever the agent operates, and carries the compliance so the platform does not have to.
Shuttle is the Payment Layer
The solution
It sits above the merchant’s existing provider, and nothing underneath changes. You integrate once, every merchant gets payment capture across voice, chat, and links, and every merchant keeps the provider they already brought, across more than forty of them.
That last part is the structural difference a competitor cannot answer with a feature: most ways of adding payments make the merchant adopt your chosen provider. The Payment Layer is the inverse.
Why carrying the compliance matters
The data behind the solution
Global average cost of a data breach in 2024, up 10% year on year.
IBM, 2024of organisations maintained full PCI DSS compliance, the last clean figure published.
Verizon, 2022 (2020 data)headless and composable platforms hold a public PCI DSS Level 1 attestation. The rest push the obligation to the merchant.
Shuttle research, 2026The Payment Layer moves that weight off the platform. PCI DSS Level 1, ISO 27001, and SOC 2 sit at the layer, and the card never enters the recording, the transcript, or the model’s context, because capture happens inside infrastructure built for it. Shuttle is Twilio’s chosen payment partner, and runs the voice payment pattern in production for AI voice platforms like PolyAI and operational commerce platforms like Brightpearl.
The project you never sign up for
of payment development never spent
in PCI compliance cost never incurred
saved per platform that stops maintaining provider integrations
What it looks like in the channel
In practice
A card captured on a live call
An inbound call reaches the agent. The conversation runs as normal until it reaches payment, and the card is captured through a secure prompt. The number never enters the recording. The transaction settles on the merchant’s own provider, and the confirmation comes back into the same call, with no transfer.
A link dropped into the thread
The agent runs the thread to the point of sale and drops a payment link into the conversation. The customer pays in place. The card never touches the transcript, and the sale lands on the merchant’s rail, inside the channel that closed it.
One identity, one transaction
A customer can start on chat and move to voice, and the cart and the card token hold across the handoff. One conversation, one identity, one transaction, however many surfaces it crossed.
What to look for
Four criteria separate a position from a patch
Bring-your-own-provider
Can each merchant keep the provider they already use?
Compliance carried, not delegated
Does it hold PCI DSS Level 1 itself and keep card data out of recordings, transcripts, and model context?
One integration, every channel
Does a single integration cover voice, chat, links, and the handoffs?
Additive, not a migration
Can it run alongside the existing web checkout without changing it?
A solution that meets all four is a payment layer. Anything that meets some of them is one of the partial answers above, and the gaps surface later, usually in the deal you most wanted to win.
Who this is for
Platforms that do not own the checkout page, and do not need to
- Vertical software running a merchant’s operations, now closing sales in the conversation
- Contact and customer-experience platforms turning calls into transactions
- ERP and finance platforms collecting payments inside workflows
- AI voice and chat agent vendors who win the customer and need to finish the sale
- Merchants who want a channel their cart platform will not give them
The page is taken. The conversation is up for grabs. The platforms and the agents that move on it now will own the next channel the way the carts own the last one.
Put the payment in the conversation
The page is taken. The conversation is up for grabs, and it is additive, so it does not wait on anyone’s roadmap. Talk to us about adding capture across voice, chat, and links on the provider your merchants already use.
Talk to us Sources
Independent research cited above
- Baymard Institute (2025). Cart Abandonment Rate Statistics: 70.22% average across 50 studies, 2006–2025.
- Gartner (2025). 70% of customer service journeys begin and resolve in conversational assistants by 2028.
- Gartner (2025). Agentic AI autonomously resolves 80% of common service issues by 2029.
- Gartner (2025). Self-service and live chat to surpass phone and email by 2027.
- Gartner (2024). 30% of Fortune 500 to offer service through a single AI-enabled channel by 2028.
- Forrester (2025). Chat-session shoppers convert at 12.3% vs 3.1%; returning chat shoppers spend ~25% more.
- Grand View Research (2024). Voice Commerce Market: $42.75B (2023) to $186.28B by 2030, 24.6% CAGR.
- Market.us (2025). Conversational Commerce Market: ~$14.9B (2024), ~15.6% CAGR; chatbots ~66%; financial services ~41%.
- Invoca (2025). Call Conversion Benchmarks: 37% of phone leads convert during the call (60M+ calls).
- PYMNTS / Restaurant Business (2024–2025). Yum Brands and Wendy’s AI voice ordering rollouts.
- McKinsey (2024). The future of AI in insurance: 10–20% conversion improvement from AI voice/chat.
- IBM (2024). Cost of a Data Breach Report: $4.88M global average, up 10% YoY.
- Verizon (2022). Payment Security Report: 43.4% of organisations maintained full PCI DSS compliance (2020 data).
- Shuttle (2026). Headless/composable platform PCI research: 4 of 16 hold public PCI DSS Level 1 attestation.
- Twilio (2024). State of Customer Engagement: companies expect to add ~4 channels; 95% maintaining or increasing.
- Zendesk (2024). CX Trends: 73% of shoppers use more than one channel per buying journey.