Parth Gaurav

So proud to be announced as finalists in @webflow professional partner of the year award category. This is for our first enterprise client @sisuclinic which we got chance to work with earlier this year. Congratulations to the whole team 🎉🚀 #webflow #NoCode

Probably the least glamorous thing I do all week: open a list of our own live URLs and just check which ones still load. Nobody really posts about this part. But a model reads what's on your page today, not the version you shipped a year ago. So honestly a lot of whether AI ever cites you comes down to boring pages that quietly still resolve, and most people never think to check.

I've been building Webflow sites for funded B2B for about 7 years, and the tell for an AI-built site has gotten really obvious this year. Even someone who's never touched a website can usually spot one in a second. With all the AI builders that flooded the market you can get a site that works in an afternoon, which is fine, but fast and trustworthy are not the same thing and buyers feel the gap. From what I've seen the AI-generated look reads as exactly that to the person you're trying to win, and for a company selling something high-trust that's kind of the worst signal you can send. What still separates a curated site from a generated one is taste, the hundred small calls about hierarchy and spacing and restraint, and that's the part AI can't really do for you yet. So the more everyone can generate a site in an afternoon, the more a considered one is worth. For a funded company the design is carrying the credibility, so honestly that's still the part I'd hand to a person.

Marketing almost never wants the dev team gone. They just want to change a headline without filing a ticket and waiting a whole sprint for it. That's kind of the entire ask. And AI can generate a full page in a minute now, which is genuinely useful. But if the build still needs a dev for every little edit, then nobody actually got their autonomy back and you've just moved the bottleneck somewhere down the line.

Half the teams asking me how AI changes their website are on WordPress. The kind of setup that breaks every time some plugin updates. AI will happily generate them a gorgeous new page. But it still sits on the same stack that goes down on a Tuesday. So honestly the tool was never the thing holding them back, it was the stack underneath that keeps breaking, and nothing you generate on top of it really changes that.

We put two new pages live for System last week. They're a public benefit corporation started by Spotify's former VP of Data, backed by Lux Capital, and they've our client for about 4+ years now. The part I keep thinking about happened after. On Thursday their team asked for two copy changes, it went to staging, and it was live inside the hour. Honestly you could get AI to make a change that fast now, so the speed was never really the point, and what came next is the part that stuck with me. On Friday a different team inside the company, one that has never worked with us directly, sent their own request into our channel. That second thing is the real proof, I think. Nobody sat down and decided we should be the default, it just became easier to send it to us than to queue it internally. If your site work still starts with a ticket and a wait, send me a DM and I'm happy to take a look at how it's put together.

AI can spit out 50 landing pages before lunch. Honestly that part is boring now, everyone has it. The thing is, what actually matters is page 51, then page 200, and whether each new one is basically free to add. If it is, someone thought hard about how the whole thing was built underneath. If it isn't you just end up with a pile of pages nobody wants to touch.

A new AI site builder shows up basically every week. Most of them won't die. They'll just find their level and sit there. And chasing each one is how you quietly stay average. The best work I've seen isn't coming from whoever has the newest tool. it's from the people who picked one thing and just kept going deeper with it. So yeah, the boring answer is usually to go deep on one thing.

A marketing team can wait two weeks to change one line on their homepage, and everyone ends up blaming the dev team for being slow. From what I've seen after about 7 years of building these sites, the dev team is almost never the actual problem. Usually it goes like this. Marketing wants to change the hero, swap the pricing, or ship a campaign page. It turns into a ticket, sits behind the product roadmap, and by the time it's live the moment's gone. Everyone assumes AI fixes this now, because you can generate a page in a few minutes. But generating a page and changing your live site without breaking it are two different things. If the build underneath needs a developer for every edit, an AI page just adds one more thing for engineering to wire in. Honestly, the teams that ship on their own didn't do it with a faster dev team, they did it with a site built so a marketer can change it safely, without touching anything fragile. So the bottleneck was never really speed, it was how the site was set up to be run. That's an architecture decision, and it gets made long before anyone's stuck waiting on a ticket.

AI can generate 70 location pages tonight. And it would, without asking a single question. But generating was never really the hard part. Knowing which 70 markets are actually worth a page, and what each one has to answer for the person who lands on it, that's the work. Someone has to own that call. A prompt won't do it for you.

The quiet milestone in a retainer is a small one. It's the first time a client loops you in before they've decided anything. No brief, no ask, just what do you think. AI can run the actual task in minutes now, that part is kind of solved. But being the person someone thinks with before the task even exists, from what I've seen that still comes from working together for a while.

We went back through our own published posts last week and found something a bit embarrassing. A study we cite a lot had a dead link. A 404. Sitting live in seven posts. Nobody flagged it. While fixing it, we found two stats attached to the wrong descriptions. Also live, also in those seven posts. Here's the part that matters. Everyone treats AEO as a launch job. Structure the page, add schema, publish, done. But content decays quietly. Links die, studies move, stats drift, and nothing tells you. If a model is deciding whether to cite you, it reads what's on your page today. Not what you shipped a year ago. So AEO is a maintenance job. Almost nobody runs it.

We went back through our own published posts last week and found something a bit embarrassing. A study we cite a lot had a dead link. A 404. Sitting live in seven posts. Nobody flagged it. While fixing it, we found two stats attached to the wrong descriptions. Also live, also in those seven posts. Here's the part that matters. Everyone treats AEO as a launch job. Structure the page, add schema, publish, done. But content decays quietly. Links die, studies move, stats drift, and nothing tells you. If a model is deciding whether to cite you, it reads what's on your page today. Not what you shipped a year ago. So AEO is a maintenance job. Almost nobody runs it.

You're not paying a developer for code anymore. AI writes clean code fine, and fast. What you're actually paying for: Someone who knows what to build in the first place. And someone who picks up the phone when the site breaks the night before a launch. The code got cheap. Judgment and ownership didn't.

The bar for a good website moved this year. Old test: does it look good. New test: does it sound like anyone's actually home. AI made every site polished, so polished stopped meaning anything. The funded companies winning right now just sound like a real team with a real point of view. Generic is the new broken.

Sisu Clinic launched 15 pages in 15 days, against a Christmas date that was not moving. They're Europe's largest dedicated aesthetic injectable provider, 25+ doctor-led clinics, backed by Greycroft. The reason that was possible has almost nothing to do with how hard anyone worked that fortnight. Their build sits on 30+ CMS collections, so a new location page or a service page is a record someone fills in, not a layout someone designs from scratch. I would say most teams get this backwards. They scope the launch, hit the date, and then every page after that costs about what the first one did. The teams still shipping quickly a year later aren't working any harder, they just built the thing once so the tenth page is cheaper than the second.

One of our funded B2B clients stopped sending me Google Docs. Now she sends an AI-generated mockup of how she's picturing the content on the page and honestly it's the best brief I've gotten. It's a stronger starting point. I can see how she's actually thinking about the layout and the hierarchy, instead of guessing from a doc. From there we do the refinement, the UX and UI calls, what actually converts, matching it to their brand. AI didn't replace the work. It made the handoff better and got us to a shared starting point faster. The mockup is a first draft of thinking, not the site. What actually converts and what a real person does on the page, that's still the job. PS: on the left is a glimpse of the mockup she sent me, on the right is the page we designed for them.

The centerpiece of the Ironflow site we just took live is a hexagon. Here's why we built it the way we did. Ironflow builds the AI operating system for defense and aerospace manufacturers. The product isn't one feature, it's a set of connected operational modules that work as a system. A single screenshot flattens that. It makes a platform look like a page. So instead of an image, we built the hexagon as an interactive, hoverable component directly in Webflow. Six modules, tessellated into one shape, each reacting on its own as you move across it. The structure of the visual matches the structure of the product. We went back and forth on this for a while. The easy version was a static diagram. The right version made the visitor feel the thing move the way the product actually does. Most people don't think you can build that in Webflow. You can. It just takes treating the site like a product, not a brochure.

I'm looking for a web designer to join our team Please click the link below, read the JD properly and fill the form tally.so/r/PdkMpV