23%
Behavior is already visible
of Americans bought something through AI in the past month, according to Morgan Stanley.
Morgan Stanley, Dec 2025Agentic commerce outlook
A clearer view of how AI agents are reshaping commerce
This page is designed for product, AI, and commerce teams that need a fast read on the category without wading through raw reports. It separates what is already happening from what is still directional.
Updated monthly from public analyst research, platform disclosures, and merchant-facing market studies.
Four signals worth tracking
Four signals worth tracking
Each signal answers a different strategic question: is the behavior real, is the flow growing, are merchants adapting, and could the channel become economically meaningful?
14x
Order flow is accelerating
growth in AI-driven orders since the start of 2025, according to Shopify.
Shopify, 202645%+
Merchant adaptation is underway
of mid-sized North American retailers already have AI tools in the customer experience.
Straits Research, 202520%
The channel could become material
of transactions could run through AI platforms or agents by 2030, in the Gartner forecast cited by Shopify.
Shopify citing Gartner, 2026Projection range by market maturity
Projection range by market maturity
Use these ranges as directional planning inputs, not as exact budgeting numbers. They help frame magnitude, not certainty.
2025
$25B-$250B
2026
$70B-$650B
2028
$250B-$2.2T
2030
$500B-$4T
Where adoption is strongest today
Where adoption is strongest today
The market is wider at discovery and narrower at autonomous execution. That is the trust gap most teams need to design around.
1
64%AI-assisted product search
2
58%Started a journey via assistant
3
41%Bought AI-recommended products
4
23%Comfortable with AI buying for them
5
4%Allow fully autonomous purchase
The adoption wave
Agentic commerce is no longer a promise: it's traffic, it's conversion, it's real money
This is the data behind why supporting multiple agent payment protocols is no longer optional. Every percentage point in this funnel represents an AI agent trying, right now, to complete a purchase in your store — using whichever protocol its platform speaks.
Monthly index · July 2026
Monthly Agentic Commerce Health
The gauge that sums up, at a glance, whether the agentic commerce wave is accelerating or cooling off — and how ready the market really is to let an agent buy unsupervised.
72/100
▲ High momentum
Adoption Pulse77.6%
This isn't the curious shopper poking at a chatbot anymore: 3 out of 4 people now open an AI before a search bar when they want to buy something. The waiting room emptied out — traffic is already migrating from the search box to the conversation.
Conversion Lift+42%
AI traffic didn't just grow +1,200% year-over-year in the 2025 holiday season — by March 2026 it converted 42% better than non-AI traffic, an all-time record according to Adobe. Every agent that bounces off your checkout because it doesn't speak its protocol is a sale that was already won.
Trust Gap62% → 4%
62% say they trust AI to shop for them. Only 4% today hand over full control of their card. That 58-point gap isn't a wall — it's pent-up demand waiting for verifiable proof that the agent acted within limits, with consent and traceability. Whoever closes that gap first captures the backlog.
Market Momentum$3–5T
McKinsey projects up to $5 trillion in agentic commerce by 2030, growing near 58% a year from today. This isn't a distant curve — it's the same slope early e-commerce had, compressed. Building the infrastructure now means arriving early to a wave that has already broken.
Does your store already speak the language these agents use to buy?
Audit your store for free →Only 23% of top retailers are fully "agent-payable" today — another 38% are one fix away. Presenc AI, 2026
Index composed by Trusteed weighting adoption, conversion, trust and growth from public data by McKinsey, Adobe, Accenture, Bain & Co. and Search Engine Land. Not an audited market measurement — a directional read, refreshed monthly.
From click to autonomous purchase: the adoption funnel
Every stage a consumer crosses is one more reason your checkout needs to speak the right protocol.
Use AI to search for products
78%
Start their shopping journey via an AI assistant
65%
Purchased an AI-recommended product
54%
Trust AI for the purchase decision
35%
Are comfortable letting AI buy for them
23%
Would delegate full purchase on small amounts (< $20)
16%
Already allow fully autonomous purchase
4%
The generation delegating today will run tomorrow's spend
% already using AI for shopping, by generation · eMarketer
Gen Z (18–27)
61%
Millennials
52%
Gen X
38%
Boomers (60+)
20%
Compared forecast (volume in USD)
*Gartner measures total agent-mediated B2B spend, not just e-commerce.
| Source | 2025 | 2030 | CAGR |
|---|---|---|---|
| McKinsey Global | ~$250B | $4T | ~58% |
| Gartner B2B (2028) | — | $15T+* | — |
| Juniper Research Global | — | $1.5T | 43.2% |
| Presenc AI Global GMV | — | $1.1T | ~27x/4a |
Why this matters for your payments stack
No single AI agent speaks one protocol: ChatGPT runs on ACP, Google backs AP2, Coinbase and Base push x402, Visa bets on VIC and Mastercard on MCAP. Consumer trust (62% declared trust, yet only 4% allow fully autonomous purchase) is growing faster than industry standardization. The #1 barrier is still data privacy (56%), not technology — meaning the infrastructure that proves provenance and verifiable consent (trust receipts, agent signatures, agent KYC) is what earns the trust missing to close that funnel.
Sources:McKinsey,Bain & Co.,Morgan Stanley,J.P. Morgan,Gartner,eMarketer,Juniper Research,Grand View Research,Accenture,Salesforce,BCG,Adobe,Presenc AI,Mordor Intelligence,Search Engine Land,Visa— data as of July 2026.
What this means for teams building now
What this means for teams building now
Design for discoverability first
If assistants cannot interpret product data, policies, and constraints cleanly, the channel underperforms before the buyer ever reaches checkout.
Treat trust as product infrastructure
The constraint is not just traffic. It is whether assistants, merchants, and buyers can understand what is allowed and when human review is required.
Avoid autonomy theater
Most teams should not optimize for blind agent execution. They should optimize for controlled delegation and faster learning loops.
How to read the data responsibly
How to read the data responsibly
Separate analyst forecasts from consumer surveys and platform signals.
Avoid treating broad AI usage claims as direct proof of autonomous transactions.
Read source dates closely. This category moves fast and mixed timelines distort judgment.
Use the combination of signals to guide strategy, not any single metric in isolation.
Source trail
Source trail
This page uses public sources already cited across the site. The goal is to keep the read fast while making verification easy.
Use the signal, not the hype
Use the signal, not the hype
If the category matters to your roadmap, the practical next step is to make your store or platform easier for agents to understand before the channel becomes more crowded.