Buy
Virtual Visa debit cards and a purchase agent that let AI agents buy online, with every charge approved by a human.
Agentcard is the most scoped payment option we've reviewed for agents that need to transact: store-locked cards plus passkey approval on every charge solve the two risks that make buyers nervous. The three-product split also matches how teams actually deploy — Vault (free up to 5,000 users, then Enterprise) if you only want to store cards, Issuing ($5,000/mo flat, $0 per card, 1% revenue share) if you need your own card program, and Purchase Agent ($0.15/request, free in beta) if you'd rather not run a merchant integration or a browser at all. It's a developer product with a real business model attached, not just a cost. Skip it if you need a physical card, or if buying from stores outside
Verified 7d ago · liveness 80/100 · cite: rightaichoice.com/tools/buy
- Agent hosting and consumer apps that want to store user cards and buy on their behalf
- B2B teams automating company purchases through a Slack or chat agent
- Crypto teams routing user funds into agent-issued cards
- Developers who want scoped, merchant-locked card spend instead of full card access
- Anyone who needs a physical card for in-store purchases
- Teams whose agents buy mostly from merchants outside the 18+ integrated stores and have no browser automation in place
- Companies that need SSO, SCIM, an SLA and a SOC 2 report on a self-serve plan — those sit behind the Enterprise tiers
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Agentcard if you need a physical card, if most of your agent's spend is at merchants outside the 18+ integrated stores and you have no browser automation, or if you need SSO, SCIM and a SOC 2 report without an Enterprise conversation.
Issuing runs $5,000/month even in a month your agents barely transact — the flat fee does not flex with volume.
The free Vault tier (up to 5,000 users) makes Agentcard cheaper than almost anything else in this category for teams validating demand, and the $0.15/request Purchase Agent is priced for occasional orders. Issuing at $5,000/month flat is expensive next to Stripe Issuing's usage-based issuing fees if your volume is low, but at high card volume the flat fee plus the 1% revenue share you keep can beat paying per card elsewhere.
In short
Buy — Virtual Visa debit cards and a purchase agent that let AI agents buy online, with every charge approved by a human. Best for Agent hosting and consumer apps that want to store user cards and buy on their behalf, B2B teams automating company purchases through a Slack or chat agent, Crypto teams routing user funds into agent-issued cards. Free to start; paid plans from $0.15.
What's new in Buy
Checked 7 days agoAcross the latest 10 updates: 10 changelog entries.
Agentcard auto-approval panel stops duplicating the shared-pool switch
Shared budget toggling now lives in one place; a pool re-enabled from that switch starts a fresh period with nothing spent.
Agentcard agent explains rejected delivery addresses
When a merchant refuses an address, the agent reports why and asks for the fix — a corrected address or recipient phone — and never edits an address unilaterally.
Agentcard reissues expired approval links
The shopping agent checks out again when asked for an approval link, replacing an expired link with a fresh one rather than returning it as valid.
Agentcard shopping agent stops rebuilding cleared carts
After a Vault approval, the agent returns the order that approval placed instead of clearing and rebuilding the cart for a second confirm.
Agentcard: partial card-network releases keep the remaining hold
Partial releases are no longer treated as full reversals; Agentcard credits only the released amount and waits for final settlement.
Agentcard: authorization adjustments now shown in wallet activity
Wallet activity shows authorization adjustments — a hold dropping from $22.97 to $20.36 now posts a Spend adjusted row with a $2.61 credit.
Agentcard blocks webhook URLs pointing at loopback addresses
Company webhook addresses aimed at loopback or internal addresses written in IPv6 form are now refused, matching existing IPv4 blocking.
Agentcard rejects processor test platforms and mismatched checkout hosts
Live payments refuse a processor's test platform so a real card is never encrypted for a test account; checkout hosts that don't match their environment are turned away.
Agentcard 3D Secure window reports the check outcome
After a bank verification, the window states how the check ended — complete, not completed, canceled, not finished, or unreadable — and offers a close button.
Agentcard supports 12-digit cards for auto-approval
Cards with 12 digits, as some networks issue, can now enable auto-approval, join the shared budget and let apps pay — previously blocked by the Vault.
What people actually say about Buy — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
62 mentions across 4 sources (Hacker News, App Store, Lemmy, Tech Press) · researched Jul 2, 2026.
Average across the 4 sources that answered — each source counts once, not each post.
- +Disposable cards protect real card details from breaches.
- +Explicit approval per transaction maintains user control.
- +Works with ChatGPT, Claude Desktop, and OpenClaw out of box.
- +CLI and MCP server enable easy developer integration.
- +Per-task card isolation prevents overcharging and misuse.
- −No community feedback available to verify real-world utility.
- −Heavy reliance on manual approval defeats full autonomy.
- −Limited merchant acceptance for one-time virtual cards.
- −Potential notification delays could stall agent workflows.
- −Pricing for Pro tier not publicly detailed yet.
Viability Score
How well maintained and how widely used is Buy? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- Vault: store customers' existing cards and buy with an agent, free up to 5,000 users
- Issuing: one-time and multi-use virtual Visa debit cards for your agent
- Cards shaped to a single merchant and a single amount
- Multi-use cards with a monthly cap for subscriptions and repeat orders
- Passkey or master-password approval before any charge is paid
- Cards encrypted on the device before they're stored
- Auto-approve within limits the customer sets
- Rewards routing picks the card that earns the most points
- Tokenback pays customers in Anthropic and OpenAI tokens instead of points
- Purchase Agent: one API call or the buy tool over MCP places the order
- Purchase Agent covers 18+ merchants including Amazon, Walmart, Target, Best Buy, Home Depot and DoorDash
- Browser integrations with Kernel, Browserbase, Browser Use and your own Playwright browser over CDP
- Messaging integrations for Linq and Blooio so iMessage agents can complete payments
- White-labeled Vault with your own identity throughout
- Funds held as USDC on Base; instant settlements
About Buy
Agentcard is a payment layer built for AI agents rather than a general card issuer with an agent wrapper. You give an agent a way to pay in one of three ways: store a customer's existing card in Agentcard's Vault and let the agent spend against it, issue one-time or multi-use virtual Visa debit cards through Issuing, or hand the job to the Purchase Agent, which places orders from your own agent with a single API call or the buy tool over MCP. Card numbers live in Agentcard's vault, so your app and the model never see the raw PAN. The controls are the point. Issued cards are shaped to a single store and a single amount; a one-time card covers one purchase and is done, while a multi-use card carries a monthly cap for subscriptions and repeat orders. Every charge is confirmed by the buyer with a passkey or master password before anything is paid. It plugs into the stack your agent already runs on: browser integrations for Kernel, Browserbase and Browser Use (or your own Playwright-over-CDP browser), messaging integrations for Linq and Blooio, and MCP for agents such as Claude and Cursor. The Purchase Agent covers 18+ merchants — Amazon, Walmart, Target, Best Buy, Home Depot, Lowe's, Macy's, Wayfair, Staples, Kohl's, B&H Photo, DoorDash, Good Eggs, Locale and flights — at $0.15 per request, free while in beta. It doubles as revenue. Issuing runs $5,000/month flat with $0 per card, a 1% revenue share on every transaction, and optional markup you set from the dashboard; funds can be held as USDC on Base, and Tokenback pays out Anthropic and OpenAI tokens instead of points. The Vault tier is separate and free up to 5,000 users. Compared with Stripe Issuing, Agentcard is narrower and agent-native rather than a general issuing toolkit.
Behind the Verdict
The reason Agentcard reads differently from most agent-payment pitches is that it treats the card as the security boundary. A card is created for one merchant and one amount, it can be single-use or capped monthly for repeat purchases, and the buyer approves with a passkey before anything settles. A leaked number buys nothing, and a disputed charge is handled like any other card dispute, insured up to $250. What's genuinely differentiated is the breadth of on-ramps. If you already run an agent browser, Agentcard attaches over Playwright CDP or through Kernel, Browserbase and Browser Use, types a placeholder card at checkout and pauses the request at the processor until the human approves. If you build on messaging, Linq and Blooio agents can complete payments inside a text thread. If you'd rather not touch merchants at all, the Purchase Agent takes one API call or the buy tool over MCP and places the order itself across 18+ merchants. The money model is unusual for the category. Issuing is a flat $5,000/month with no per-card, per-user or per-order fees, a 1% revenue share on settled volume paid back to you, and markup you can change from the dashboard without a deploy. You can hold funds as USDC on Base and pay customers in Anthropic or OpenAI tokens via Tokenback. The Vault tier is free through 5,000 users, which makes it cheap to validate demand before committing to an issuing program. The honest weaknesses: the product is early-stage, and the Enterprise tiers are the only place you get SSO/SCIM, a named contact, an SLA in writing and a SOC 2 report — companies that need those on day one are pushed to a sales conversation. Physical cards aren't part of this. Store coverage is real but finite: the merchant list runs through Amazon, Walmart, Target, Best Buy, Home Depot, Lowe's, Macy's, Wayfair, Staples, Kohl's, B&H Photo, DoorDash, Good Eggs, Locale and flights, and buying elsewhere means bringing your own browser automation. And $5,000/month on Issuing is a real commitment for a pre-revenue product, even with the flat-fee math working in your favor at volume.
Researching Buy? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Buy actually fits — and what changes day-one when you adopt it.
You put a user's existing card into the Vault, then let your agent check out at a supported store; the user gets a passkey prompt on their phone before the charge lands.
Outcome: The card number never touches your app or the model, the user approves every charge, and disputes are handled like any other card dispute with coverage up to $250.
You embed the Purchase Agent with one API call from your Slack agent, confirm the exact cart, and let it buy from Amazon, Walmart or Staples without writing merchant integrations or running a browser.
Outcome: Orders go live in an afternoon, users approve with a passkey, a changed price comes back to you before purchase, and every order reports its order number, delivery window and shipments.
You hold program funds as USDC on Base, issue cards to users in nearly any country, and let their agents spend while you take a 1% revenue share plus markup you set from the dashboard.
Outcome: Cards are funded from crypto, settlements are instant, and the revenue share plus markup turns the payment layer into a line of business rather than a cost centre.
Use Cases
- Store a customer's card in the Vault and let an agent buy on their behalf without the app ever seeing the PAN.
- Issue a single-use virtual card locked to one merchant and one amount for a one-off agent purchase.
- Give a subscription-running agent a multi-use card with a monthly cap instead of full card access.
- Embed the Purchase Agent in your own agent with one API call and order from Amazon, Walmart or DoorDash.
- Let an iMessage or WhatsApp agent built on Linq or Blooio complete a payment inside the thread.
- Attach Agentcard to a Kernel, Browserbase or Browser Use session so your browser agent can check out.
- Earn interchange: take a 1% revenue share on settled volume and set your own markup from the dashboard.
- Hold program funds as USDC on Base for a crypto-native agent product.
Models Under the Hood
as of 2026-10-04
Limitations
- Issuing starts at $5,000/month, and Enterprise tiers for Vault and Issuing (SSO/SCIM, named support contact, written SLA, SOC 2 report and pen-test summary) are sales-negotiated rather than self-serve.
- Purchase Agent merchant coverage is a defined set of online stores (Amazon, Walmart, Target, Best Buy, Home Depot, DoorDash and others); other merchants require running your own browser agent against an issued card.
- Vault is free only up to 5,000 users.
as of 2026-10-01
Verification history
We have re-verified Buy 7 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 7 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Buy tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Vault Free
$0/mo
Ideal for
Consumer or B2B apps with fewer than 5,000 users that want to store customers' cards and let an agent buy without touching the card number.
What this tier adds
Free entry point: card storage from any country, passkey approval, auto-approve limits, rewards routing, white-labelling and an insured-against-mistakes guarantee.
Vault Enterprise
Custom
Ideal for
Companies past the 5,000-user cap that need SSO, SCIM, a written SLA and a SOC 2 security review before storing cards at scale.
What this tier adds
Adds unlimited users, ownership of the Vault and every card in it, a named contact with a shared Slack channel, SSO/SCIM, custom approval rules and a custom contract.
Issuing
$5,000/mo
Ideal for
Agent products with real transaction volume that want their own card program, a revenue share and markup instead of paying per card.
What this tier adds
$5,000/month flat with $0 per card issued or connected, a 1% revenue share on every transaction, optional markup, Tokenback rewards and funds held as USDC on Base.
Issuing Enterprise
Custom
Ideal for
High-volume issuers that need volume-based rates, SSO/SCIM, a named contact and a security-reviewed custom contract.
What this tier adds
Adds volume pricing that comes down with card volume on top of every Issuing feature, plus dedicated support, SLA, SSO/SCIM, custom limits and a security review.
Purchase Agent
$0.15/request (free in beta)
Ideal for
Developers who want agent purchases across 18+ merchants without building merchant integrations or running their own browser.
What this tier adds
Starting tier for buying: $0.15 per request (free in beta), one API call or the buy tool over MCP, passkey approval from the user's Vault card and order tracking.
Where the pricing makes sense
The company stage and team size where Buy's pricing actually pencils out — and where peers do it cheaper.
The free Vault tier (up to 5,000 users) makes Agentcard cheaper than almost anything else in this category for teams validating demand, and the $0.15/request Purchase Agent is priced for occasional orders. Issuing at $5,000/month flat is expensive next to Stripe Issuing's usage-based issuing fees if your volume is low, but at high card volume the flat fee plus the 1% revenue share you keep can beat paying per card elsewhere.
Setup time & first value
How long it actually takes to get something useful out of Buy — broken out by persona, not the marketing-page minute.
Vault: minutes — store a card and auto-approve limits are set in the dashboard. Purchase Agent: live in an afternoon, one API call or the buy tool over MCP, with no merchant integrations and no browser to run. Browser integration: the Kernel, Browserbase, Browser Use and CDP snippets are copy-paste, so a working checkout is typically an afternoon. Issuing: production in under a week with no sales
Switching to or from Buy
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From Stripe Issuing: move card creation to Agentcard's Issuing API and keep the merchant-locked, single-amount card model for agent spend.
- →From a shared company card in an agent's environment: store it in the Vault instead and gate each charge behind passkey approval.
- →From a hand-rolled Playwright checkout: attach Agentcard over CDP and let it hold the payment at the processor while the human approves.
- →From manual order placement: route the request through the Purchase Agent and confirm the cart before it buys.
- →From a points-only rewards program: switch to Tokenback and pay customers in Anthropic or OpenAI tokens.
- ↗To Stripe Issuing: rebuild card creation and controls on a general issuing toolkit if you need payment rails beyond agent purchases.
- ↗To a raw browser-automation stack with no card-issuing layer: keep your Playwright code and handle card storage and approvals yourself.
- ↗To a messaging-native payments provider: if your product is entirely inside iMessage or WhatsApp, drop the card layer and let the platform handle the payment.
- ↗To manual procurement: remove the agent from the payment path and approve purchases by hand if the volume doesn't justify the program fee.
Integrations
Resources & Guides
Tutorials & Learning
YouTube returned 6 videos for “Buy”, and we withheld 6: 6 could not be judged, because “Buy” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Buy.
Official links
Featured Head-to-Head Comparisons
Buy vs Bitsgap
Buy and Bitsgap serve entirely different domains. If you need to let AI agents spend money safely with disposable virtual cards and per-transaction approval, Buy is your only practical option. If you want automated crypto trading bots across multiple exchanges with backtesting and demo mode, Bitsgap is the clear choice. They are not competitors but tools for distinct use cases.
Buy vs Truleo
Buy and Truleo serve entirely different markets: Buy (Agentcard) is a payment infrastructure for developers building autonomous AI agents, while Truleo is an intelligence platform for law enforcement. There is no direct competition. Choose Buy if you need to let AI agents make purchases safely; choose Truleo if you need to unify police data and automate lead generation. Your decision depends solely on which job you need done.
Buy vs Presto Voice
Buy and Presto Voice are not direct competitors; Buy is a payment infrastructure for AI agents, while Presto Voice is a drive-thru voice AI solution for QSR chains. Your choice depends on whether you need to let AI agents spend money autonomously (Buy) or automate drive-thru ordering to boost revenue (Presto Voice). For developers building agentic workflows, Buy is the clear pick; for restaurant chains wanting to reduce labor costs and increase upsells, Presto Voice is the established leader.
Popular in Agent Memory & Runtimes
Frequently Asked Questions
Used Buy? Help shape our editorial sentiment research.