How AI Is Changing the Creative Process Inside Agencies
The tools made it easy to produce a hundred ideas by lunchtime. They did not make it any easier to know which one is actually good.
Two years ago, an agency pitching a client would spend days building a mood board and a rough storyboard by hand before a single frame of the actual film existed. Today, a junior creative can generate a passable moodboard, a rough animatic, and three headline variants before lunch, using tools that did not exist when most of the senior people running the pitch learned the craft. The tools changed fast. What the industry still has not agreed on is what a creative person is actually for once the tools can do this much of the work alone.
The parts of the job that got faster, not better
AI has been genuinely useful for the unglamorous, time-consuming parts of agency work: generating quick visual references, drafting multiple headline options to react to, mocking up a rough cut of a script before committing budget to a real shoot. This has compressed timelines that used to take days into hours, which is unambiguously valuable for pitch speed and client iteration. What it has not reliably done is generate the single sharp idea that a campaign is actually built around, the kind of insight that comes from a strategist noticing something true and specific about how people actually behave, which remains stubbornly human work.
Taste became the scarce skill, not output
When generating options becomes nearly free, the actual bottleneck in the creative process shifts to judgement: knowing which of fifty AI-generated variations is actually worth pursuing, and why. This has quietly changed what agencies are paying for. A junior creative who can produce a large volume of competent-looking work is now less valuable than someone with the taste and experience to look at that volume and correctly identify the one option worth developing further, a skill that AI tools do not meaningfully replace and that takes years to build, not a subscription to shortcut.
The tools made it easy to produce a hundred ideas by lunchtime. They did not make it any easier to know which one is actually good.
The trust problem nobody has fully solved
Clients in India have been notably cautious about AI-generated content appearing in finished, paid advertising, partly over authenticity concerns and partly over unresolved questions about copyright and the provenance of training data behind these tools. This has pushed AI's role in most serious agency work toward the internal, unseen parts of the process, research, drafting, mockups, rather than the final customer-facing execution, at least for now. That line is moving, but slower than the pace of the tools themselves, because brand reputation risk moves at a much more cautious speed than software development does.
What the craft actually needs to hold onto
The agencies handling this transition well are not the ones resisting the tools, nor the ones outsourcing judgement to them entirely. They are the ones using AI to compress the mechanical, repetitive parts of the process while investing more deliberately in the genuinely human parts: strategic insight, cultural fluency, and the taste to know a good idea from a merely competent one when both are sitting on the same screen. The technology solved the problem of generating options quickly. It has not touched the much older, much harder problem of knowing which option deserves to exist in the world, and that remains the actual job.
Related on Ad Tribe
Advertisement
