The loudest take regarding AI and web agencies right now is that agentic development has made building a site such a trivial endeavor that digital agencies are next up on the proverbial chopping block. I don't see it that way, and a recent conversation with Jesse Nieman, Head of Sales at Finsweet, gave me a clearer way to explain why.

Jesse and I have known each other for years, mostly as friendly competitors in the Webflow space. Our call was part of an ongoing series, Building The Next Web, that has also included conversations with Neel Shivdasani from Webflow, StackedGTM’s Josh Grant, OFF+BRAND’s Stu Ross, and more, and I wanted to use this one to talk through what agentic development is doing to agency work, versus what the internet thinks it's doing.

Why agentic development makes agencies more necessary

Jesse told me about a project his team scoped after a founder had already vibe-coded most of a marketing site himself. It looked good until the founder tried to make one update and the whole thing fell apart. That story isn't rare anymore, and the Edgar Allan team has seen the same pattern: someone gets to 80% in a weekend, and the last 20% is where everything they didn't know they needed to know comes knocking on the door.

This isn’t anything new, really. The real constraint has always been the unresolved stuff underneath the build: what the company stands for, who it's for, how the story holds together across a hundred pages, how a marketing team works together to ship and govern content at scale.

So the reality is, agentic tools solved the fastest, least valuable part of the web development process, but they don’t touch the part clients are really paying for.

The build has always been the easy part

I said something on the call that I've said to clients for years, and that made me want to say it again in writing: by the time we're building, we've already answered the hard questions. What's the company's story? What's the market position? Who needs to agree on the direction internally before anything ships? Getting a leadership team aligned on a mission they believe in is a much harder problem than assembling a page.

A one-shot AI build sets the floor for what a website can do. The floor is a company that’s online. The ceiling is a site that holds up under an AEO audit, survives a rebrand, and keeps earning a buyer's trust long after launch.

That baseline is a genuinely great one, better than what most small businesses could access a few years ago. But reaching higher still requires upstream processes like story engineering and the wisdom work that most agencies skip past to get to the “building” part.

“Both/and” beats “either/or” when picking a tech stack

Jesse made a point I think the whole industry needs to hear: Webflow hasn't gotten worse in the last six months. It just stopped being the only option, and a lot of people are confusing "there's a new alternative" with "the old thing failed." Finsweet is still building most client work in Webflow while also shipping Next.js and React projects when a client's problem calls for it. And that’s the right instinct. You want your loyalty to be to the client’s outcome, not the platform. That’s just a means to getting there.

Edgar Allan operates the same way. Our stack includes Webflow, Profound, and our own ag-nts data, but the honest answer to "what should we build this in" is almost always "it depends on what you're trying to solve." 

Curiosity closes more deals than pitches do

Here’s part of the conversation I'd want every account lead at Edgar Allan to hear: Jesse said his team wins deals by asking more questions than the competition does. He goes into sales calls trying to help a prospect understand their own project, sometimes surfacing questions they hadn't thought to ask themselves yet. If someone leaves a call thinking "I don't know much about this agency yet, but they clearly understand my problem," that's the win. The formal proposal comes after.

That’s a sales philosophy that assumes the agency's job starts with listening, starting with a brand’s signal in the digital world (which we assess through a purpose-built Claude skill), and then extends to how we see the creation of content that drives brand visibility. All before we ever talk about a fix or an editorial strategy.

The agency’s job has shifted from building sites to growing them

The other shift worth naming: Edgar Allan is seeing this firsthand. Clients aren't hiring us just to build a functional or beautiful website anymore. They're hiring us to build one that keeps growing after launch. On our recent accounts, AEO, A/B testing, and on-page personalization have stopped being upsells and started showing up as default scope, even for smaller brands that might not have thought beyond launch a few years ago.

The aperture of "web work" has widened to include everything from structured content to how a brand shows up when someone asks an AI model about it, and clients increasingly expect one partner to hold all of it instead of stitching together a website team, an SEO vendor, and a content shop that don't talk to each other.

That's the case for what we call Visibility Engineering and Optimization: brand, build, and AEO working as one connected system instead of three separate contracts. A brand that's designed, built, and optimized in the same place is coherent in a way assembled work rarely is, and that coherence is what makes it legible to AI systems and to customers alike.

If you're evaluating agencies right now, ask them how they'd solve your problem before you tell them what you think the deliverable should be. The ones asking good questions back are the ones worth a second call. And if you’d like us to take a crack at answering your questions, contact us.