Nathan Huening, Principal Solutions Engineer on Webflow's enterprise team, gave us a clear answer to the question every marketing leader has been asking this year: Is Webflow dying?
The answer? Webflow isn't going anywhere. What's going away is the way most people have been building websites for the last decade.

So, every few years, a platform debate gets hot and eventually catches fire into a referendum on the entire category. This year's version has three things burning at once: WordPress was supposed to kill agencies, no-code was supposed to make developers obsolete, and now AI code generation is supposed to be obliterating Webflow.

None of those other things has happened yet, by the way, and there's no good reason this round will be any different. What's really going on is there’s a rift forming in the POV around what a website is for, and it's drastically reshaping which agencies are thriving and which are stuck.

I put the headline question to Nathan as part of Edgar Allan’s Building the Next Web series, which exists to ask the people building the internet's infrastructure what's changing and what isn't. It includes conversations with Finsweet’s Jesse Nieman, StackedGTM’s Josh Grant, and Island’s Ari Yablok. And this one, friends, was a good one.

There are two kinds of websites, and only one of them needs a strategic partner

Nathan draws a pretty clean line between what he sees as the two kinds of websites, and it’s the key to the whole debate.

Some sites are brochures. A restaurant's menu, a law firm's list of practice areas, a manufacturer's contact page. Take one down for an afternoon, and not a lot happens: the business hardly blinks, customers just pull the phone number off Google Maps instead. These kinds of sites are everywhere, and they're perfectly fine as they are. They just don't really need the firepower of a professional platform like Webflow, an agency, or a deep content strategy. Squarespace quite literally does the job, and that’s OK. Nobody needs to overthink it.

Then there are sites where the business lives inside the website: class signups routed to front-desk software, franchise data syncing across dozens of locations, a marketing stack of 30 or 40 tools knitted together with a CMS as one cog in the machine. Knock that site offline for two hours and revenue dips. This is where enterprise clients exist, and it's a category where a website build never really ends.

Most “is X dead” debates confuse the issue because they lump these two categories into one market. They're not the same. One needs a vendor you call when something breaks. The other needs a strategic partner who knows the business well enough to keep improving it long after launch.

The real reason enterprise Webflow agencies can't hire fast enough

Solo builders and small studios doing one-off sites are watching demand dry up. Meanwhile, enterprise-focused Webflow agencies can't staff fast enough to keep up with the work coming in. And it comes down to business model, not company size.

A one-time build puts an agency on a treadmill: spend months learning a client's business, ship the site, then go find the next client and start the whole thing over. A managed enterprise relationship runs on different math. You spend three to six months on a site the client will live inside for the next decade. SEO, accessibility, AEO, and performance aren't problems you solve once. They're maintenance you do every week, and most companies aren't set up to do that maintenance themselves.

An enterprise website is never a finished product. It's an asset that either gets tended weekly or slowly stops working, and most of a company's marketing budget depends on which one is happening.

That's the shift we’ve built our whole model around at Edgar Allan. We don't hand a client a finished site and then disappear. And that longer relationship isn’t about locking a brand into a restrictive retainer; it’s conceived as a feature of the type of work we’re doing. Our Customer Success Program exists specifically for the work that starts after launch: the testing, the content, and the ongoing optimization that determines whether a $10 million ad budget, for example, converts.

What Nathan's grocery agent proves about the future of websites

Nathan built his own AI agent to do his grocery shopping. Feed it a recipe URL, and it maps every ingredient to a SKU at his local store, then checks out on its own through an MCP integration. On the surface, that looks like a website getting cut clean out of the picture, one more point for the "it's all dying" argument.

But look closer, and it's the opposite. 

For that agent to turn "two cups of flour" into an actual product and check out without a human double-checking the order, the store's product data had to be structured clean enough for a machine to act on it unsupervised. That's the exact job a storefront has always done: represent the business accurately enough for a transaction to go through. The interface changed. The requirement for clean, governed data behind it didn't budge, and neither did the fact that somebody has to keep that data clean.

Which is exactly the kind of site Nathan was describing; the kind that never really finishes. A grocery store's SKU data isn't a one-time build any more than Orange Theory Fitness's front-desk integration is. Nathan points to Orange Theory as the enterprise version of the same problem: a class signup has to route from the site into front-desk software in real time, correctly, threaded through a marketing stack of 30 to 40 connected tools. Whether it's a recipe mapped to a grocery SKU or a class booking mapped to a franchise location, both sit firmly in the category of “things that need tending,” not the category of “needs a template.” The businesses winning in this moment already know this, and build their data to be acted on by a machine in the first place.

The website's new job: a source of truth for agents and humans

None of this changes what the right response to a zero-click future is. It's still not spending less on the website. It's building a more resilient one.

A website used to have two jobs: capture an email, capture a credit card. Now there's a third, and it belongs to the same category of ongoing work as everything else in this piece: govern what the brand says about itself everywhere else, including inside the AI models increasingly standing between a customer and the decision to buy.

Some content should stay wide open so AI systems can read it and cite it correctly. Some belongs behind a form, or behind a real conversation with a real person. Getting that split right takes the same discipline as any brand strategy work: knowing exactly what you want to be known for, and saying it consistently enough that a model can repeat it back without garbling it. That's the whole premise behind our own Visibility Engineering and Optimization methodology, which treats brand clarity, SEO, AEO, and CRO as one connected system instead of four separate line items.

Most agencies work one lane of this problem and call it done. Some do brand. Some run AEO audits. Some build in Webflow and stop there. But a brand that's articulated, built, and optimized under one roof stays coherent in a way assembled-from-parts work almost never does. That coherence is exactly what an AI system needs to describe you correctly, and exactly what keeps a site resilient while the tools around it keep changing.

Side note: If you want to see where your own site has gaps, our brand clarity assessment is a solid place to start, and it pairs well with the diagnostic work in our AEO playbook.

The tools will keep changing shape. Whether a site was built to survive that, clear enough about who it is that a machine can repeat it back accurately, and tended closely enough to keep being true, won't.