There's a conversation happening in agency pitches right now, and it goes something like this: a client looks at the scope, looks at the timeline, and says "but if AI made it faster, shouldn't it be cheaper?" It's a reasonable question. It's also the wrong one. And how an agency answers it will determine whether they survive the next three years.
The received wisdom
The assumption baked into that question is that price tracks time. You paid for hours. AI reduced the hours. Therefore: discount. It's a tidy logical chain. Accountants love it. Procurement teams love it. And it's quietly destroying the positioning of every agency that accepts it. Here's what that logic misses entirely: the thing you were actually buying was never the hours. It was the outcome. The strategy behind the campaign. The judgement call on which creative direction won't backfire. The accountability when the tracking breaks. The person who's done this thirty times before and knows which shortcut will cost you three times as much in six months. None of that got cheaper. If anything, in a world where AI floods the zone with fast-produced content, the judgement layer got more valuable.
What actually happened when AI arrived
The agencies that are winning right now didn't just swap human hours for AI hours and pocket the margin. They reinvested. The same time that used to go on a first-draft brief now goes on a second round of creative testing. The production hours saved on a landing page go back into a properly built variant framework. The speed gain unlocks something that was previously unaffordable: iteration. This is what founders who stay close to their work understand instinctively. Efficiency isn't the product. Better output is the product. The best agencies are using AI to produce more proof, run more tests, move faster through the learning loop , and then they're pricing on the results that loop produces, not the hours it consumed. The data backs this up. A retainer-model agency running AI-assisted delivery is carrying roughly 8 points more net margin than an hourly equivalent. Not because they're charging more per hour. Because they've moved off the clock entirely. The price is the outcome, and the outcome doesn't change based on how many keystrokes it took.
The discount demand is a positioning problem, not a pricing problem
When a client pushes back with "AI made it faster, so it should be cheaper" , that's not a negotiation problem. It's a symptom. It means they see your service as a time-and-materials contract dressed up in strategy language. The price pressure is just where that misunderstanding surfaces. The fix isn't a better counter-argument in the pitch. It's earlier. It's in how you position the work from the first conversation. If you're talking about hours, deliverables, and process, you'll get negotiated on hours, deliverables, and process. If you're talking about outcomes , pipeline generated, conversion rate improved, cost per acquisition reduced , you get evaluated on those, and AI efficiency becomes your competitive edge, not your liability. The agencies that handle this best treat AI tooling the same way they treat Slack, their project management software, their QA process: infrastructure. It's in the overhead. It's how they maintain standards. It's not a pass-through cost that fluctuates with how clever the tools got this quarter.
What this means for your business right now
If you're buying agency services: the question to ask isn't "why hasn't the price dropped?" It's "what are you doing with the efficiency gain?" A sharp answer , more variants tested, faster iteration, better measurement , tells you they're investing it back into your results. A vague answer about "streamlined process" tells you they've pocketed it and cut corners. If you're running an agency: the window for this conversation is now. Clients are forming these expectations. The ones who accept the discount framing first will find it almost impossible to reverse. The ones who define value on outcomes, lock in retainer structures, and show the proof of what the efficiency buys , those are the ones building something defensible. Speed without quality is just getting to the wrong answer faster. AI gives you speed. What you do with it is still a human decision.
The close: AI didn't make the judgement cheaper. It just made the cost of getting it wrong higher.
X POST (max 260 chars)
Most agencies asked "AI made it faster , shouldn't it be cheaper?" and said yes. That's the trap. The efficiency gain isn't the discount. It's the extra iteration, the variant tests, the proof you couldn't afford before. New piece on the AI pricing trap 👇 [link]
LINKEDIN TEASER
There's a question appearing in almost every agency pitch right now: "AI made production faster , so why hasn't the price dropped?" It's the wrong question. And the agencies accepting the discount framing are making a mistake they'll struggle to undo. The efficiency gain isn't the client's to claim. It's the fuel for better output, faster iteration, and the kind of proof that previously didn't fit in the budget. Price the outcome. Invest the speed. Read the full piece → [link]
Draft by Sage 🎨 | 2026-07-03 | Foundry Works blog pipeline Word count: ~680 words Status: AWAITING JASON APPROVAL
FAQ
Should agency services get cheaper because AI makes production faster?
Not automatically. If the agency is using the efficiency to run better testing, produce stronger proof and improve outcomes, the value of the work can increase.
What is the AI efficiency trap?
The AI efficiency trap is the belief that faster production should reduce price, even when the real value is strategy, judgement, iteration, accountability and commercial outcome.
How should agencies answer AI discount pressure?
Agencies should move the conversation away from hours and toward outcomes: pipeline generated, conversion improved, cost per acquisition reduced, proof produced and learning cycles accelerated.
What should clients ask instead of asking for an AI discount?
Ask what the agency does with the saved time. Strong answers include more variants, faster iteration, better measurement and clearer evidence of what is working.