Ploy.ai, the Y Combinator company founded by former Webflow co-founder and CTO Bryant Chou, sells a seductive promise: paste a URL, wait about a minute, and watch an engine called the “slurper” ingest your live site “down to the last pixel.”
The marketing is frictionless. The migration-cost calculator frames agencies as the expensive alternative and Ploy as the free, instant one. “Ploy slurps your site into an editable, optimized version — URLs, structure, and SEO preserved — in minutes, for free.” Chou has described the slurper as a deterministic system that cost roughly $750,000 in compute to build, reconstructing components and a design system in about 75 seconds.
That last sentence is the tell. The product is not a mover. It is a reconstructor. It does not lift your site off the open web and hand you files you can host anywhere. It rebuilds a Ploy site — on Ploy hosting, edited by Ploy agents, metered in Ploy credits, with structured content steered into PloyDB, and with source access gated behind a paid plan. The slurp is the on-ramp. The garden is the destination.
This is not a personal attack on Chou, who spent twelve years building Webflow’s product, CMS, and hosting. It is an argument about incentives and architecture. When a vendor makes entry a one-click party trick and exit a paid, partial, time-boxed afterthought, the business model is lock-in — whether or not anyone uses the word.
What “web slurp” actually does
Ploy’s own language is more honest than the migration framing. A slurp is “a best-effort, deterministic (no-AI) pass that pulls your brand’s soul — structure, identity, design tokens.” Credits then turn that extraction into “professional-grade Astro.” The company is explicit: “A slurp isn’t a 100% clone.”
Official docs are equally clear about the clone’s limits. Site cloning captures rendered HTML at import time. Dynamic content — CMS-driven pages, user-generated content, API data — comes across as static. Collections and live connections need separate setup. Private pages are not guaranteed.
That matters because Ploy’s growth story does not stop at a prettier static copy. The recommended next step is to move structured content from WordPress, Webflow, Squarespace, Framer, Wix, or Sanity into PloyDB, then reconnect the pages that publish it. Notion databases get the same treatment.
So the funnel is:
- Slurp the public surface of your site.
- Flatten dynamic systems into a snapshot.
- Re-hydrate that snapshot inside PloyDB.
- Point the domain at Ploy (or put Ploy in front of the domain with a reverse proxy).
- Let agents keep rewriting, researching, and publishing on Ploy.
Chou has described the intended end state without much euphemism: slurp first, then “push any new page live on its own, or migrate your entire site onto Ploy.” Hybrid hosting is offered as a courtesy, not as the product thesis. You can keep Webflow or Framer behind a fallback origin and route selected paths to Ploy. Then, when you are ready, “cut the proxy when the last page is migrated.”
That is not a coexistence strategy. It is a staged annexation of your hostname.
Why a website is a uniquely dangerous thing to put in a garden
A website is not a slide deck. It is the public identity of the company: URLs that other people have linked, schema that search engines and answer engines cache, forms that feed the CRM, content models that encode how the business talks about itself, analytics history that tells you what worked, and brand components that every campaign inherits.
Vendor lock-in is not “you pay a subscription.” Switching costs are normal. Lock-in is when the vendor amplifies those costs on purpose: proprietary runtimes, incomplete exports, integrations that only work inside the suite, and a content model that does not map to anything else. Lynton’s “Five Locks” framework names the pattern — code, content, data, process, and commercial — and the code lock is the oldest trick in marketing platforms. Templates written in HubL, Liquid, or a vendor visual language have a market value of zero the day you leave.
Closed website builders already ran this experiment. Wix has no full-site export; layout, animations, and structure live in the editor. Squarespace will give you a partial content export, not the design. Leaving either platform is a rebuild, not a transfer. Agencies who do this work say it plainly: on those platforms you rent the site; you do not own it.
Webflow — Chou’s previous company — is the more sophisticated version of the same problem. Paid plans can export HTML, CSS, JavaScript, and assets. CMS collections, memberships, ecommerce, form handling, search, and password protection do not come with the ZIP. Dynamic collection data does not travel as a working system. You can leave with a static snapshot of the look. You cannot leave with the machine.
Ploy is that lesson applied to an agent era. The slurp compresses the expensive part of getting in. Nothing in the public product story compresses the expensive part of getting out.
The six locks inside Ploy
1. Source code is a paid privilege
Ploy’s FAQ answers the portability question directly:
“Your full site source (React, Astro, and Tailwind) is available through Code Sync, which connects your site to a GitHub repository you control. Code Sync is included with every paid plan. Free workspaces can’t access the underlying codebase.”
Read that again. The free slurp — the thing the homepage, the startup page, the enterprise page, and the migration calculator all push — does not include the right to take your code. You can ingest a site in 60 seconds without a card. You cannot take the reconstruction home unless you start paying.
That is not an accident of early-stage docs. It is the classic freemium ratchet: make the demo irreversible, meter the escape hatch.
Even the paid hatch is not “download a ZIP and leave.” It is GitHub sync into a repo that still has to satisfy Ploy’s “publish contract” if you want the platform to keep deploying it. The CLI’s first-class workflows are bring-your-repo-into-Ploy, Code Sync with Ploy workspaces, and moving a site between Ploy workspaces — not exporting a self-contained production system that no longer needs Ploy’s agent, PloyDB access keys, or Ploy hosting.
Astro, React, and Tailwind are real, portable technologies. That is better than Wix’s black box. It is not ownership. A repo that still expects Ploy’s runtime, environment secrets, database access keys, routing layer, and agent-authored component conventions is a souvenir, not an exit.
2. The domain becomes their front door
Custom domains are how a marketing site becomes a business asset. Ploy’s own domain docs set plan allowances by “domain family”:
- Free: 0 custom domain families
- Starter ($50/mo): 1
- Pro ($300/mo): 5
- Enterprise: unlimited
Ploy does not take over nameservers. It asks you to point A and CNAME records at Ploy, then provisions SSL.
The reverse-proxy feature is more subtle, and more important. Point the domain at Ploy, keep the old origin as fallback, and route /blog or /pricing onto Ploy pages “until they hit a path you’ve routed to a Ploy site.” Marketing copy calls this “no big-bang migration.” The same page describes the endgame: migrate section by section, then cut the proxy.
Once DNS points at Ploy, Ploy is no longer a vendor you evaluate. It is the router for your public hostname. Uptime, TLS, redirects, robots, sitemaps, and which origin serves which path become Ploy problems. That is operational gravity. Teams stop treating the old CMS as the source of truth because the address bar already lives elsewhere.
Chou has recommended this path in public: slurp a HubSpot site, improve selected pages, “or have us host your whole domain and reverse proxy the pages you don’t want to migrate.”
3. PloyDB is the new CMS lock
Ploy’s enterprise pitch is not “we host pages.” It is “one platform replaces seven contracts.” The first line item they propose to replace is the CMS itself — Webflow, Contentful, or a headless system, priced in their table at $30k–$200k a year — with “native PloyDB-backed content.” Visitor identification, AEO, analytics, page generation, and continuous optimization are bundled the same way.
PloyDB is “the database built into Ploy.” It powers structured website content, CMS workflows, directories, forms, and other records “without a separate database or CMS account.” Docs recommend database-backed blogs as the default: posts as rows, templates as Ploy-built pages, live edits without a republish, plus an access key the site needs to read the posts. File-based content exists as a lesser option. External CMS connections exist. The recommended path is inward.
This is the content lock. CSV export of rows, if it exists in the CLI, is not the same as exporting a working editorial system: draft states, relations, scheduled publishes, form inboxes, generated collection routes, and the agent workflows that know the schema. SaaS CMS history is full of teams who discovered that an export format does not map cleanly to anything else, and that a “two-week migration” became six.
Every post, case study, changelog entry, and comparison page you let agents write into PloyDB is inventory that is cheapest to keep there.
4. Credits turn maintenance into a meter
Ploy does not primarily sell a site plan. It sells credits.
Free is 2,000 credits (800/day cap). Starter is $50/mo for 4,000. Pro is $300/mo for 24,000. Image generation, code generation, document generation, planning, visitor enrichment, and Ploybook runs — including scheduled ones — draw from the same balance. Chatting with the agent to edit the site spends credits. Heavier “professional” builds spend more. Direct text edits in preview usually do not; AI fallback for text that cannot be edited directly does.
The product vision is a site that “works while you sleep”: overnight rewrites of high-bounce pages, scheduled Ploybooks, continuous AEO and SEO passes. That vision is also a usage pump. A website that is “self-improving” is a website you cannot cheaply freeze. If agents are the way the site is maintained, then leaving Ploy is not just leaving a host. It is firing the only staff that knows how the site was made.
This is pricing lock-in of a modern kind. You are not trapped by a three-year enterprise contract. You are trapped because the operational muscle of the site — the thing Chou calls a “company brain” — is a metered API sitting on top of your public property.
5. The suite is the switching cost
Ploy’s homepage and enterprise pages do not sell a website builder. They sell consolidation: builder + analytics + visitor identification + CRM sync + AEO citation tracking + ads + CRO experiments. “Stop stitching marketing tools together.” “One platform three engines.” “Seven contracts fewer.”
That pitch is commercially rational. It is also how lock-in compounds.
- Analytics that only exist because they are “built-in” and server-side on Ploy hosting do not come with you as a product. They come with you, at best, as an export of events.
- Visitor identification that resolves U.S. traffic to companies and pushes accounts into Attio or HubSpot is valuable and vendor-shaped. The graph of who visited what lives in Ploy.
- AEO tracking across ChatGPT, Perplexity, Gemini, Claude, and AI Overviews is a dataset you will not want to orphan.
- Forms, lead routing, and Ploybooks that draft outreach from identified visitors are process lock: the team’s weekly rhythm assumes the agent.
Gartner-era SaaS advice is boring because it is true. Once data has been converted and upstream and downstream apps have been implemented around one vendor, switching is “quite complicated and costly.” Vendor selection is a high-risk decision because of that.
Ploy is designed to become that vendor for the entire front office of a startup. The earlier you “plug Ploy into your business,” YC copy says, “the smarter it gets about what actually works for you.” That sentence is a feature if you never leave. It is the definition of path dependence if you do.
6. The contract is a 30-day box on the way out
Ploy’s Terms of Service, as indexed from the company’s legal page, follow the standard SaaS pattern with one clause that should be taped to the procurement checklist:
You retain rights in Customer Data and in AI outputs generated from that data, subject to the Terms. The Services themselves — software, models, algorithms, interfaces, documentation, branding — remain Ploy’s exclusive property. Feedback you give is assigned to Ploy. Either party may terminate. Upon termination: the license ends immediately; Ploy will make Customer Data available for export for thirty (30) days, after which it may be deleted; fees owed become due.
The privacy policy describes a sixty-day post-termination retention window for export before deletion or de-identification. Even the more generous number is not a migration program. It is a countdown.
“You own your content” is the sentence every walled garden prints. The operative questions are always: own it in what format, with what relationships intact, on what timeline, and does the running system come with it? A 30-day export window after you have pointed production DNS at a vendor, filled PloyDB, and trained the team on Ploybooks is not portability. It is a courtesy.
SOC 2 Type II, as of the enterprise page, was still “in progress.” That is normal for a seed-stage company. It is not a reason to make them the system of record for the company website.
The Webflow irony is the point
Chou is not a stranger to this tradeoff. He helped build a platform that now, by his own accounting, powers on the order of 1% to 1.5% of the internet. Webflow’s success was real. So were the complaints that followed: CMS that does not export as a working system, hosting that is hard to leave, a visual language that does not travel. Those complaints are documented in Webflow’s own help center, not just in competitor blogs.
Ploy is the sequel with better onboarding. The first company made it easy to build inside a garden. The second makes it easy to ingest the open web into a garden. The slurper exists because the expensive part of lock-in used to be the rebuild in. Remove that cost, and the only cost that remains is the rebuild out — which the customer has not felt yet.
There is a reason the migration calculator estimates agency rebuilds at $2,500–$8,000 for a brochure site, $8,000–$25,000 for a mid-size site, and $25,000–$75,000+ for a large content or ecommerce property, then contrasts that with a free slurp. Those agency numbers are also a rough map of what it will cost to leave Ploy once PloyDB, custom routing, agent-written pages, and native analytics are the production system. The calculator is showing you the exit fee. It is just labeling it as the price of not coming in.
After $27 million in seed funding led by First Round and Y Combinator, the company is not incentivized to make you a happy short-term tenant. It is incentivized to become the operating system for the website and everything that touches it. “The earlier you plug Ploy into your business, the smarter it gets” is a retention strategy with a research-and-development budget.
“But they use Astro” is not an escape hatch
This is the strongest objection, and it deserves a straight answer.
Ploy sites are described as Astro + React + Tailwind, with a CLI, GitHub Code Sync on paid plans, environment variables, and even a path to bring an existing Astro repo in if it satisfies a publish contract. That is more adult than Squarespace. It is closer to the AI-app-builder class (Lovable, v0, Bolt) that at least pretends the repo is yours.
Portability still fails in layers:
The page layer. Agent-authored components, design-system conventions, and Ploy-specific wiring are not a public framework with a second implementation. You can fork the files. You inherit a codebase nobody on your team wrote and that was optimized to be edited by Ploy’s agent.
The data layer. PloyDB access keys, form records, and CMS rows are the living site. Static Astro collections were an earlier/partial story; the recommended path is the hosted database. File-based content is documented as the less capable option.
The traffic layer. Custom domains, route rules, reverse proxies, fallback origins, and SSL are Ploy-hosted. Cut the account and you cut the hostname unless you have already rebuilt DNS and hosting somewhere else.
The growth layer. Visitor graphs, AEO citation logs, experiment history, and scheduled Ploybooks do not compile to static files.
The labor layer. If the team’s skill is “ask the agent,” the team cannot maintain the export. That is the quiet lock-in of agent-native products. The artifact is code. The competence lives in the vendor.
Owning a GitHub repo is step one of about six. The rest of the steps are what make a website a business system rather than a folder of components.
What you actually risk
SEO and AEO debt. A slurp can preserve URLs on the way in. A sloppy exit will not. Titles, schema, sitemaps, internal links, and answer-engine citations are easy to brag about as “automatic” on Ploy and easy to break when you leave. Ploy’s own calculator lists those items as agency line items. They become your line items on the way out.
Brand entropy. The slurper’s selling point is consistency: one component system, no AI slop. That system is Ploy’s reconstruction of your brand, not your design team’s source of truth. Over time the “source of truth” is whatever the agent last shipped. Figma integrations do not change that if production is the agent.
Operational concentration. Website, CMS, analytics, identity resolution, outreach drafts, and ad landing pages in one seed-stage company is a single point of failure. Startups accept that trade for speed. They should not pretend it is architecture.
Negotiation leverage. Once production DNS, content, and weekly growth rituals sit in one product, price increases, credit-pack changes, feature gates, and plan reshuffles are not theoretical. SaaS history is the same story on a loop: generous entry, then the bill for the thing you cannot extract. Contentful’s free-tier and pricing shifts in the mid-2020s were a reminder that “we’ll always be cheap to start” is not a covenant.
The second rebuild. Teams leave Wix and Squarespace when they outgrow templates, hit performance ceilings, need custom backends, or get tired of renting. They pay again. Ploy’s pitch is that agents postpone that day. They also raise the height of the wall you will eventually climb, because the site will have more pages, more programmatic SEO, more account-specific landing pages, and more Ploy-native data than the brochure you slurped.
The honest use cases — and the rule that keeps them honest
There are non-stupid ways to touch this product. They all share one rule: Ploy may generate. Ploy may not own the system of record.
Reasonable:
- Slurp a copy of a marketing page into a sandbox to steal a first draft, then rebuild in a repo you already deploy (Astro, Next, Webflow, WordPress, whatever you control).
- Use Ploy as a throwaway landing-page factory for ads, with the canonical site still on infrastructure you can export tomorrow.
- Use the reverse proxy only as a temporary experiment, with a written kill date, while DNS still can fall back without Ploy.
- Pay for Code Sync on day one if you insist on building there, and treat the GitHub repo as the source of truth — not the agent thread.
Unreasonable, if you care about optionality:
- Point the primary domain at Ploy as the long-term origin.
- Migrate the CMS of record into PloyDB.
- Let scheduled Ploybooks be the only people who know how pages get made.
- Keep the only copy of analytics and visitor identity inside Ploy.
- Accept a free slurp and assume you can “just export later.”
If a vendor’s best feature is how fast they can swallow an existing property, ask how fast they can spit it back out as a running site. Ploy’s public answer is: pay us, connect GitHub, hope the publish contract and the database story are enough, and if you cancel, you have thirty days.
That is not a migration tool. That is a capture tool with excellent taste.
What “owning your website” actually means in 2026
Owning a site is not a vibe and it is not a terms-of-service sentence about Customer Data. It is a test you can run without asking the vendor’s sales engineer.
- Can a competent developer who does not work at the vendor stand up the production site on another host this month?
- Does the CMS export as structured data and as a working template system, or only as a pile of rows?
- Do forms, auth, search, and experiments survive without the vendor’s APIs?
- Is the design system in a format other tools speak (tokens, standard CSS, public components), or only in the vendor’s agent memory?
- If the company disappeared on a Friday, would Monday’s homepage still resolve?
WordPress survives mockery because the files and the database can be copied. Headless CMS products that speak ordinary APIs survive because content can move even when the presentation layer is rebuilt. Static generators survive because the output is files. Closed builders fail the test, and then they sell you a calculator that prices the failure as someone else’s problem.
Ploy’s slurp is a remarkable piece of engineering. That is not in dispute. The trap is the product wrapping: an all-in-one “growth engine” that wants to be the website, the CMS, the analyst, the CRO intern, and the front door of the company, funded to make that dependency deepen every week you stay.
Paste a URL if you want to see the parlor trick. Do not confuse the trick with a strategy. A business that lives on the web should be able to leave the gardener. If the only way out is another slurp — this time of their garden, by the next vendor with a $750,000 crawler — you were never migrating. You were being passed from enclosure to enclosure, and calling it growth.
Citations:
- Y Combinator — Ploy company page
https://www.ycombinator.com/companies/ploy - Ploy — homepage
https://ploy.ai/ - Ploy — Website Migration Cost Calculator
https://ploy.ai/migration-cost-calculator - BigGo Finance — “Bryant Chou Built an AI That Redesigns Websites in 75 Seconds…”
https://finance.biggo.com/news/ef6407cb7236ec10 - Ploy Support Docs — home
https://docs.ploy.ai/ - YouTube — “Why Ploy.ai is more than another website tool feat. Bryant Chou”
https://www.youtube.com/watch?v=V__2_3_l4_g - Ploy — AI Marketing Platform for Startups
https://ploy.ai/for/startups - Ploy — Enterprise
https://ploy.ai/for/enterprise - Ploy — Hosting / reverse proxy
https://ploy.ai/features/hosting - Ploy — Analytics
https://ploy.ai/features/analytics - Ploy — Visitor identification
https://ploy.ai/features/visitor-identification - Ploy — Answer engine optimization
https://ploy.ai/features/answer-engine-optimization - PR Newswire — “Ploy Raises $27M…”
https://www.prnewswire.com/news-releases/ploy-raises-27m-to-turn-your-companys-website-into-your-hardest-working-employee-302803231.html - Ploy blog — $27M announcement
https://ploy.ai/blog/ploy-raises-27m - TechTimes — “Webflow’s Bryant Chou Bets AI Favors the 40-Year-Old Founder…”
https://www.techtimes.com/articles/319406/20260630/webflows-bryant-chou-bets-ai-favors-40-year-old-founder-ploy-raises-27-million.htm - Ploy — Pricing
https://ploy.ai/pricing - Ploy — Terms of Service
https://ploy.ai/terms-of-service - Ploy — Privacy Policy
https://ploy.ai/privacy - Ploy Docs — Credits
https://docs.ploy.ai/credits - Ploy Docs — FAQ & Troubleshooting (Code Sync / export)
https://docs.ploy.ai/faq - Ploy Docs — Ploy CLI
https://docs.ploy.ai/cli - Ploy Docs — Custom Domain Setup
https://docs.ploy.ai/custom-domains - Ploy Docs — CMS & Blogging / PloyDB
https://docs.ploy.ai/cms - Ploy Docs — Security & Compliance
https://docs.ploy.ai/security - Ploy — Ploy vs. the Fragmented Marketing Stack
https://ploy.ai/marketing-stack - Ploy — Ploybooks (including “Migrate a CMS into PloyDB”)
https://ploy.ai/ploybooks - GitHub — Ploy-AI/ploy-tools
https://github.com/Ploy-AI/ploy-tools - Bryant Chou on X — slurp / migrate thread
https://x.com/bryantchou/status/2070516575527215304 - Bryant Chou on X — HubSpot slurp + reverse proxy
https://x.com/bryantchou/status/2068924212937666902 - Webflow Help Center — “How do I export my Webflow site code?”
https://help.webflow.com/hc/en-us/articles/33961386739347-How-do-I-export-my-Webflow-site-code - Lynton Library — “The Five Locks: How SaaS Vendors Keep You Trapped”
https://www.lyntonweb.com/library/the-five-locks/ - Unfold CMS — “Self-Hosted CMS vs SaaS CMS”
https://www.unfoldcms.com/blog/self-hosted-cms-vs-saas-cms/ - TechTarget — “Challenges of implementing SaaS successfully”
https://www.techtarget.com/it-infrastructure/tip/Challenges-of-implementing-SaaS-successfully - Graphic ReDesign — “The Wix and Squarespace lock-in…”
https://graphicredesign.com/blog/wix-squarespace-lock-in - Break the Image — “WordPress vs Squarespace vs Wix”
https://breaktheimage.com/resources/wordpress-vs-squarespace-vs-wix - Michalak — “Webflow and Framer: The Vendor Lock-In Dilemma”
https://michalak.world/posts/webflow-and-framer-the-vendor-lock-in-dilemma/