We left Framer (and others will, too) — Frigade

LoginAssistantManage your AI assistantEngageBuild onboarding flows

Get Started

Products<br>Assistant<br>Engage<br>Company<br>How It Works<br>Pricing<br>Updates<br>Blog<br>About

LoginGet Started

← All writing<br>We left Framer (and others will, too)<br>Code, then Framer for two years, then back to code. Why we made the move, and why most teams running a marketing site on a no-code tool will too.<br>Eric Brownrout, Co-founder<br>·May 11, 2026·5 min read

Frigade.com has lived three lives: hand-built in code during YC, two years on Framer, and now back to code. We rebuilt the third one in 267 commits this past week, voice-only, no IDE, talking to Claude.

The speed of the rebuild matters because speed was Framer's whole pitch. Their billboard a few blocks from our office reads "Enterprise needs. Startup speeds." We just rebuilt our entire site faster than an agency could quote a merge of the two we were already running on Framer.

Framer's pitch, on the way to the airport. Exactly what we couldn't get out of the tool.No-code tools like Framer and Webflow were built as abstractions over code for humans managing complexity. LLMs don't need those abstractions. The audience these tools were designed for has changed, and the trade-offs are about to change for almost everyone else running a marketing site on a no-code platform.

This post is the why part of the story. If you came for the how, that's in our previous post.

Why we went to Framer in the first place

Frigade.com had two lives before this rebuild. We hand-built the first version on Next.js in our YC days, when there were three of us and we wanted the site to feel like a piece of product. It worked for over a year. Then it stopped working.

The code wasn't the problem. The cost of running it was, and the cost was paid in engineer time. Every change pulled someone, almost always me, off product work. With three or four people on the team, "we have ten things to ship this week and the marketing website is one of them" almost never came out in the website's favor. We'd ship a bare-bones update and get back to product. Scrolling through old versions of frigade.com from late 2023, you can tell which weeks we'd been busiest and were phoning in the site updates.

This is the dynamic Framer and Webflow were built around. Engineering time is too valuable to spend on a marketing site, so put it in a designer-friendly tool and hand it to the marketing team or a contractor. Free your engineers up for product. The whole no-code premise relies on it.

So we hired our first agency and moved to Framer. They leveled up the brand and shipped a beautiful site for Engage. The honeymoon was real. The day an agency hands a site back is the day the site is at its peak.

The first Framer site, built for Engage.Why Framer stopped being the right call

The tax came back, in four forms.

Any third-party agency sits outside the customer and product context. Every brief turned into a painstaking review cycle to pull the work back toward what we'd actually shipped.

The agency held the brand. We could touch copy and tweak layouts, but new illustrations, custom components, or animations needed their hand.

New page templates outside what they'd already built us cost a few hundred to thousands of dollars each.

Framer charges per seat. We didn't add one for everyone on the team, and ramping up on a separate tool for occasional edits wasn't worth it anyway. Edits queued up on me, and when I was heads-down on product, the site froze.

So the site stagnated. We stopped touching it for stretches.

Then we got ready to launch Assistant, and we didn't see how Assistant fit into the existing marketing site. We hired a new agency, more on-brand for the kind of product work we wanted for that launch. They built us a slick second Framer site on a fresh domain at the time (frigade.ai) with a lot of custom Rive animation. It was beautiful.

The second Framer site, built for the Assistant launch on frigade.ai.But now we had two Framer sites, the same four-fold tax doubled, and an incoherent customer experience. Merging them inside Framer would have meant weeks-to-months of agency work and tens of thousands of dollars. The reasons we'd left code in the first place, slowness and quality, were happening to us inside Framer.

We tried hard to stay. The simplest test: could I tell Framer's AI "build me a new on-brand case study page using our design style" and get something I could ship? The answer was seemingly no. If it had been yes, this post wouldn't exist. The tooling can do small edits well enough; what it can't do is compose components and brand language from a brief, which is the thing that actually matters.

We have a lot of respect for what Framer built. Their taste is impeccable, and the two years we spent on the platform were the right call when we made it. None of this is a knock. The trade-offs just changed.

How fast it actually was to leave

Once I started the spike, the...