Babou.ai
Agents made your dev 10x, Babou does it for your marketing.
We left Framer because agents could not drive it deeply enough.
We picked Framer for the obvious reason: design and code in one place. But once we tried to build our actual marketing site with agents in the loop, the workflow broke down. Linked components, hover interactions, variants, SVGs, and component internals all required manual editor work or exposed opaque hashed IDs.
So we moved the site to Next.js, Tailwind, and MDX.
The surprising part: design got faster. Instead of nudging boxes, we generate variants, review them, iterate, and ship. The design surface became something agents could actually operate on.
My takeaway: if a tool is part of an operational workflow in 2026, "can an agent drive every meaningful part of it?" matters more than "does it have a nice visual editor?"
Full writeup:
2 Comments
This post is your positioning, and it is sharper than anything on babou.ai. "If a tool is part of an operational workflow in 2026, can an agent drive every meaningful part of it matters more than does it have a nice visual editor" is a better headline than your homepage has right now. You went looking for a Framer story and wrote your own manifesto by accident. The hashed IDs, the opaque component internals, the manual editor work: that is the whole reason a product like yours needs to exist, told as a lived story instead of a claim.
The deeper cut, and the thing worth pulling out explicitly, is that the tool was not really the problem. The unit of work was. Framer made you manipulate: nudge a box, wire a variant, click into internals. Next plus MDX let you generate: produce variants, review, ship. Design did not get faster because you changed editors, it got faster because you stopped manipulating and started reviewing. Agents do not want a nicer canvas, they want a surface where the job is generate-and-approve. Name that shift, because it is the real one and almost nobody says it out loud.
And that is exactly what Babou is for marketing: product goes in, variants come out, a human reviews and ships, the same move you just made for your own site. Which makes this Framer post the most convincing top-of-funnel you have, because it shows the belief working on your own product before you ask anyone to buy it. One question: does the full writeup end by connecting "we left Framer" to "so this is the belief Babou is built on," or does it stop at the dev story? Because that bridge is the sale you are one sentence away from making, and it would not read as a pitch, it would read as the reason you exist.
Once agents start touching real marketing workflows, the hidden friction usually shows up in the parts that were never meant to be automated.
A lot of systems still assume humans will handle the “last 10%”, which is often where iteration actually happens.
What part of the design-to-ship loop do agents currently fail at most for you?