Every few months, a new wave of anxiety sweeps through design Twitter. The prompt-to-image tools get better. The component generators get smarter. Someone posts a Figma file built entirely with AI and the replies split between "we're done" and "this is soulless garbage." Both are wrong. Both are also missing the point.
The question isn't whether AI can design. It can — with the right prompts, the right tools, and a human who knows what good looks like. The question nobody is asking is: what does that do to the economics of the agency model?
The Agency Model Was Already Broken
Traditional agencies run on labour arbitrage. You hire mid-level designers and developers, mark up their time, and sell that blended rate to clients who don't know what they're actually paying for. A $25,000 website might represent three weeks of real work, padded into a 12-week timeline to justify the billing structure.
This worked because clients had no reference point. They didn't know how long things should take. They didn't know what was actually hard and what was administrative overhead. They paid for the reassurance of a process, not the output of work.
AI doesn't just speed up design work. It exposes exactly how much of a traditional agency engagement was never really about the work at all.
When a founder can prototype a landing page in an afternoon using v0, Framer AI, or Cursor — and it looks good — the gap between what agencies charge and what the work costs becomes visible. That's a pricing crisis. And pricing crises kill industries.
What AI Actually Compresses
AI tools have gotten brutally good at specific things:
- Component generation — UI elements that used to take hours to build from scratch now take minutes to scaffold and refine.
- Copy iteration — Headline testing, variant generation, and tone adjustment that used to require a copywriter for every round now happen in seconds.
- Asset production — Icon sets, illustrations, and placeholder imagery that agencies used to source or commission.
- Code translation — Converting designs to production-ready code, a step that historically doubled timelines.
These aren't the interesting parts of design work. They're the parts agencies billed heavily for because they were time-intensive, not because they were skilled. Strip those out, and the bill for a typical agency project should drop by 40–60%.
Most agencies are pretending this isn't happening.
What AI Cannot Compress
Here's where the "AI will replace designers" narrative falls apart. The parts of design that AI genuinely cannot replicate are the parts that actually matter:
Strategic framing
Understanding why a product exists, who it's for, and what it needs to say to that person at the moment they land on it — this requires business acumen, not pattern matching. AI will confidently produce beautiful work in the wrong direction. Knowing the right direction requires a human who understands the problem deeply.
Taste and restraint
AI is trained on what exists. It optimises toward the aesthetic centre of its training data. That produces competent, derivative work — which is exactly what most clients don't need. The value of a skilled designer is knowing what not to do. Knowing when to strip something out. Knowing when a constraint makes something better. AI has no instinct for restraint.
Accountability
Someone has to be responsible for the outcome. When the website launches and conversions are flat, someone has to diagnose why, make a call, and fix it. AI is a tool. Tools don't own outcomes. Designers do.
What This Actually Means
Studios that use AI well aren't replacing designers — they're changing the ratio. One senior designer with good AI tooling can produce what three mid-level designers used to produce. That's not job destruction at the level of craft; it's restructuring at the level of the team.
For agencies, this is existential. Their entire cost structure — the junior team, the account managers, the project managers, the revision rounds — relied on work taking a long time. Compress the timeline and you compress the justification for the overhead.
For clients, it means what a quality digital product costs is going to drop significantly over the next two years. Not because quality is getting cheaper, but because the inefficiency tax is being removed. The studios that survive will be the ones that were already building on skill and taste — not on billable hours.
We built RYC on this premise. Small team, high AI leverage, senior judgment on everything. The output quality is higher than most agencies twice our size. The timelines are a fraction. That's not a marketing claim — it's just what happens when you stop charging for the parts of the process that AI has made nearly free.
The agencies that are going to struggle aren't the ones who build with AI. They're the ones who keep pretending they don't need to.