PlatformDTC for enterprise
ConnectClaude Codeand it builds your store.
Original research
We measured how 147 DTC brands look to a shopping agent
The DTC AI Crawler Index requests every brand’s discovery documents and records what an agent can actually read. The full method, the raw results and the two numbers we had to correct are all published.
Free tool
See your product page the way a crawler does
The agent readiness checker fetches your page with JavaScript disabled, reads the structured data an agent would parse, and tells you what is missing. No account required.
Study
Agentic checkout is already live on half of DTC
We called the endpoints on 58 well-known DTC hosts. 31 serve a live UCP discovery document and every one of them answers a Model Context Protocol tools list. The hosts and the transcript are in the post.
Security
The control plane that makes autonomous agents safe to run
Scoped keys, an approval gate on spend, idempotent mutations and a full audit trail — the Agent Gateway security model, documented end to end.
Total cost of ownership
Compare what your current platform actually costs to run
Platform fee, payment margin, app subscriptions and the engineering time replatforming consumes — laid out side by side against what the same operation costs here.
Migration
Run PlatformDTC in parallel before you cut over
Import your catalog, rebuild the storefront against your brand and keep the old platform live until the numbers agree. Nothing about the switch has to be a leap.
Solutions
One system for how you sell
Online, B2B and in-person orders share one catalog and one order record.
Why PlatformDTC
One system beats a stack of apps
Every app in a stack keeps its own copy of your customer. Here there is one: the order, the subscription, the dispute and the parcel are the same record, for your team and for your agent.
Built by operators
Our founder ran a DTC business earning $20M a year in profit. We fix the breaks we paid for ourselves.
One order, the whole story
Your agent reads one order and gets its fraud risk, any open dispute, the tracking number and the ad click it came from, in a single call.
First-party where money moves
Checkout, subscriptions, fraud, support and tracking are built by us, not rented from apps. The App Store is for the long tail.
Maximize revenue
Checkout runs on your own domain, with Apple Pay, Google Pay and PayPal built in.
Peak-load performance
Checkout is served from the edge: median time to first byte 46ms and largest contentful paint 180ms in our own Playwright runs. Live availability per service is published on the status page.
The difference
Fifteen apps keep fifteen copies of your customer
One record, nothing to sync
Refunds, disputes, renewals and tracking read and write the same order, so nothing drifts out of sync and no report has to reconcile two apps.
Your agent works on the real thing
Connect Claude, ChatGPT, Claude Code, Codex or Cursor through one MCP server, with a scoped key per agent and every write logged.
Checkout built for conversion
Checkout is served from the edge, with wallets and express payment methods built in — see the checkout details.
Instant, immersive global storefronts
Storefronts and checkout are served from a global CDN, close to wherever your customers are rather than from one origin region.
Developer-grade tools, built to extend
Comprehensive docs and flexible APIs for the teams building on top.
One order lifecycle
Storefront, subscriptions, payments, inventory, returns and fulfillment on a single system
100 governed agent operations
Scoped dtc_* tools and idempotent writes
Built for subscription brands doing $1M to $50M a year
Run PlatformDTC alongside your current stack until the numbers agree, then move. Tell us what you run today and we will plan the switch with you.
Frequently asked questions
- When is replatforming actually justified at enterprise scale?
- When the constraint is structural rather than a matter of effort. Four situations qualify: subscription or fulfilment logic that has outgrown what an app-integration boundary can express reliably, per-order platform and app cost that has become a real margin line, needing agents to perform write operations under scoped and audited credentials, or a capability that is genuinely unavailable rather than merely awkward. Slow themes and app sprawl are implementation problems that survive a migration intact.
- What is the integration-boundary problem?
- When subscriptions or fulfilment run as a separate system synced to commerce, the two disagree constantly in small ways: inventory that does not reserve against upcoming renewals so you oversell, discounts evaluated by two engines that stack differently, analytics that cannot attribute a renewal to the acquisition that produced the original order. Each is individually minor and collectively is where a great deal of operational time goes. It is invisible during a demo.
- What should we establish before signing anything?
- Three facts, all knowable in a week and all decisive. Whether stored payment credentials can move, in writing, from both processors — if not, every active subscriber re-enters a card and you lose a share permanently. How many URLs have earned traffic or links in the past year and whether the new structure has a one-to-one equivalent for each. And whether subscription state — next bill date, plan, discount history — reconciles, because errors there surface as wrong charges rather than failed imports.
- Why is onboarding sales-led rather than self-serve?
- Because the three facts above decide the outcome and none of them survives a signup flow. A migration that goes wrong on redirects, credentials or subscription state is worse for the brand than not migrating at all, and no amount of product polish compensates for getting them wrong.