Opinion
The Real AI Win Is in the Boring Parts
Image generation gets the attention. The bigger near-term gain for a studio or a shop is handing the repetitive, unbillable work to assistants like GPT, Claude, and Manus, and getting your hours back for craft and clients.
The loudest argument about AI and creative work is about images: who made them, who owns them, whether they are any good. It is a real argument. It is also not where most shops will actually win this year. The quieter, larger gain is hiding in the parts of the day nobody bills for: the rambling client call that has to become a brief, the proposal that has to get written, the spec that arrives as a mess, the follow-up email at nine at night. That is where assistants like GPT, Claude, and Manus earn their keep, and it has almost nothing to do with making art.
It helps to know what you are working with, because there are two kinds. Conversational assistants, like OpenAI's GPT and Anthropic's Claude, are strong at language and reasoning. They draft, summarize, restructure, explain, and help you think out loud. Autonomous agents, like Manus, go a step further: you hand them a goal and they plan and carry out the steps, browsing, analyzing, assembling a document or a simple page, and hand back a finished draft rather than a reply. One is a sharp assistant in a conversation; the other is closer to an intern who goes off and does the task. Both exist for the same reason, which is leverage.
The practical uses pile up fast once you start looking. Turn a forty-minute client call into a clean, structured brief and scope. Draft the proposal, the change-order note, and the diplomatic message explaining that a supplied file is too low-resolution to print well. Generate a list of forty name or tagline options to react to, instead of staring at a blank page. Write alt text, product descriptions, and search-friendly captions for a whole campaign in one pass. Summarize a twelve-page request for proposal down to the actual asks and dates. Pressure-test a pitch by asking for the strongest objection a client could raise. Research substrates, finishes, or a regulation quickly, then verify what you find. Stand up a small quote calculator or a production checklist. None of it is glamorous. All of it is time, and time is the one thing a busy shop never has enough of.
The catch is that leverage cuts both ways, so a few rules keep it from cutting you. Verify everything; these tools are confident even when they are wrong, and an agent, as one reviewer put it, is an impressive intern you still look over the shoulder of. You own the output, mistakes included. Mind confidentiality; do not paste a client's sensitive material into a tool without knowing how that data is handled and whether you are permitted to. Keep your voice; the default tone of these models is recognizable, and if every message you send sounds like a machine, clients will notice, so edit it back into your own words. And do not outsource the thinking. The goal is to remove friction, not judgment.
Used that way, this is leverage aimed exactly where the industry is heading. The solo owner who used to lose every evening to admin can run lean and spend the reclaimed hours on the work that is actually theirs. The two-person studio can pitch like a five-person one. It is the same story as accessible equipment and a decentralized market: capability that used to require a payroll is now within reach of one person with a vision. The assistant does not make you less of a craftsperson. It clears the underbrush so you can spend more of your day being one.
The flashy promise of AI is that it will make your art. The useful promise, the one you can bank this quarter, is that it will clear your desk. Let it draft the email, untangle the brief, assemble the spec, and run the errand. Then take the hours it hands back and put them where they were always meant to go: on the work only you can do, and the clients only you can keep.
About this piece. This is an opinion piece on creative workflow and tools; it is not legal or security advice. The capabilities, pricing, and data-handling practices of AI tools vary and change, so confirm a tool's current terms and your own confidentiality obligations before using it on client work.