Point of View · Creative Leadership
How I lead a creative team in the age of AI.
For years, when a team needed to move fast, I threw gasoline on the fire and brought in outside designers. Now the gasoline is AI. Here is how I actually direct it, and the tools I would put in my team's hands.
For most of my career, the thing standing between a good idea and great work was rarely the idea. It was everything around it.
Sometimes it was production. You have the concept, you just need forty banner ads by Friday. Sometimes it was talent, and every creative director knows the feeling of finally hiring the designer who reads your mind, the one who takes half a sentence of direction and hands back something better than you pictured. And very often it was systems. When a brand has its templates built, its guidelines set, and its component library in order, shipping a campaign is almost easy. What image are we using? What is the headline? The button color is already decided. When none of that exists, every single asset turns into a fresh argument.
AI can be all three of those things. A production engine. A second set of hands. The system nobody ever got around to building. But only if you direct it. Left on its own, AI is the intern who hands you fifty options that are all a little bit wrong, and now you are the one wading through lukewarm work trying to find the one usable idea in the pile. That is slower than doing it yourself.
The magic is on the front end. You write the brief. You bring the original idea, the one that came out of your own head. Then you hand AI something real to work from. Give it a blank prompt and you get blank work. Give it your thinking, your voice, and your guardrails, and it becomes the fastest, most tireless person on your team.
Give it nothing and you get nothing back. Give it your thinking and it becomes the fastest hand on the team.
That is the whole job now. Not making the work. Directing the making of it.
Where I am coming from
I have spent years in SaaS, inside tech companies with genuinely great talent. When we needed to move fast, when we needed to throw gasoline on the fire, I brought in outside designers and agencies to expand what my team could do in a week.
AI is that gasoline now. It is the surge capacity I used to go out and buy, except it lives inside the team full time, and it only works as well as the person holding it. I am also an entrepreneur. I run my own business, I set the strategy, and I still do the work myself. So I say this from both sides of the desk: I have used AI to save myself entire days, and I have used AI in ways that cost me hours and left me completely frustrated. The tool was never the difference. The direction was. This is how I make sure it lands on the right side of that line, for me and for a team.
The line I draw
Once this line is clear, most of the daily questions answer themselves. The short version: AI is brilliant at polishing an idea you already have, and useless at having the idea for you.
Where I let AI run
The time eating work. This is where it buys back real hours.
- Competitive analysis and opportunity scans
- Research on what competitors have already done
- Audience and persona exploration
- Proofing and pressure testing my work
- Checking every link and fact before a post goes live
- Reformatting, resizing, and localizing approved work
- Summarizing feedback, notes, and long transcripts
Where a human starts
The parts that carry the idea, the voice, and your name.
- The original idea, always from a person
- The strategy and the creative brief
- The first draft of the copy that matters
- The brand voice and the final quality bar
- Taste, the final cut, the call on what ships
- Rights, ethics, and accountability
Notice the first draft sits on the human side. Put nothing in and you get crap out. Even a moodboard comes back sharper when you start it with a point of view.
How I run it
Still write the creative brief.
Do not stop writing briefs just because AI can spit one out. The brief is where the strategy and the original thinking live. Write it the way you always have, then hand it to everyone, your people and your tools alike, so the whole team is working from one source of truth. Same brief in, aligned work out.
The human brings the idea. AI polishes it.
This is the order that matters, and I see it done backwards constantly. A person inputs the idea, the angle, the concept. AI helps refine it, tighten it, and check it. Then a person decides. Flip that order, let AI generate and a human pick from the pile, and you get average work chosen quickly. The good stuff still starts in a human head.
Protect the brand voice with living guidelines.
Your brand guidelines should be a living document that evolves as you evolve, including as you bring AI in. Load them into your tools and update them when the brand shifts. Ten people prompting off ten different mental models will produce ten different brands. One living source keeps everyone sounding like you.
A great idea does not need twenty options.
I do not believe in exploring for the sake of exploring. The ideas, the copy, the visuals, the actual creative genius, that all starts with a person. Once you have the idea, AI earns its place: polish it, sanity check it, run it against the brief and the guidelines. But sometimes you already have the answer, and you do not need AI handing you nineteen weaker versions of it. Use it to sharpen, not to stall.
Keep a human name on everything that ships.
Someone owns the final call, the facts, and the tone of every asset. If it goes out under our brand, a person stood behind it. Accountability does not get automated.
Stay legal, and get smart about AI as the new front door.
Confidential and customer data stay out of public models, full stop, and that goes double for anything unreleased like a product launch. Make sure the team has the right permissions and tools in place before anyone touches sensitive work. At the same time, there is a right way to feed AI. These chatbots are becoming a new kind of search engine, and your SEO team can be putting accurate information about your company into AI search the correct way, so your brand shows up as the answer. It is a whole new discipline and it is genuinely fascinating. Just do it the right way.
Manage each person the way they need to be managed.
This is a philosophy I lead by, and AI is no exception. Sit down with each of your direct reports and build an AI plan for that person. What tools do they actually need? Where are their weak spots? Where are they losing hours to mindless work a machine could take off their plate? A copywriter and a senior designer will use AI completely differently. So do not roll out one blanket policy and walk away. Talk to each person, make a plan built for them, and keep adjusting it. You are the creative director. Leading this is the job.
The stack
I get asked what to actually use, so here is the lineup I would build. Most of these earn their keep in time saved. Tools move fast, so treat this as my current starting roster, not gospel, and note that everything below still runs through the line I drew above.
Everyone, day one
One shared assistant, loaded with your brand
Before anything role specific, I get the whole team onto one AI assistant and load the brand guidelines, the voice, and our best examples into it, so every person is drafting from the same source of truth. I lean toward Claude for this, because its Team plan and Projects hold a large amount of brand context and tend to follow it closely, which keeps writing on voice. ChatGPT with a shared Custom GPT does the same job and is a little easier to pass around. Pick one, set it up once, and make it the front door for the team.
Copywriters
The assistant is the tool. The voice is still theirs.
Their shared assistant (Claude or ChatGPT, loaded with the brand) does the heavy lifting: drafting from a brief they wrote, tightening, repurposing one piece into ten, and checking facts and links before anything publishes. It is their editor and research partner, not their ghostwriter. The idea and the final voice stay with them.
Designers
Concept fast, produce faster
Adobe Firefly and Photoshop generative fill and expand for commercially safe image work right inside the workflow. Midjourney for concepting, moodboards, and high quality reference once there is a clear direction to point it at. Figma AI for fast first layouts and content fill. Canva for quick, templated social when speed matters more than polish.
Video, fast
Where the time savings are wild right now
Google Veo for polished, cinematic brand moments. Runway for controlled image to video and real editing when a producer is driving. Kling for fast, high volume concept and hook testing, the UGC style stuff you want ten quick versions of. Descript to edit video as easily as a doc and pull social clips out of longer footage. Use these for concepting, social variants, and mood, not for exact product demos or regulated claims, and put a human eye on every frame.
Social, fast
One brief becomes a week of content
Once the brief exists, the shared assistant is genuinely great at social copy in the brand voice. Pair it with a repurposing tool like Descript or Opus to turn one video into a week of clips, and a scheduler like Later or Buffer for captions and planning.
Training the team
Two things make or break this. The first is that training is per person, not per company. Go back to that AI plan for each report, get them genuinely confident on the two or three tools that fit their role, and let them ignore the rest. A designer does not need the copy stack. A writer does not need Runway.
The second is the standard, and I am firm on it. If you are my copywriter and you are handing me AI output that I have to edit back down to something usable, that is not acceptable. You are the copywriter. If you had written it from scratch it would carry your stamp and your name, and it still has to. Using AI as a tool means figuring out how it actually makes your work better, not how it lets you pass off a rough draft and call it done. I have felt the frustration of AI that wastes your time firsthand. The people who win with it are the ones who put in real direction and own what comes out the other side.
Green light, let AI run
- Research, competitive scans, and opportunity analysis
- Proofing, fact checking, and link checking before publish
- Reformatting, resizing, and localizing approved work
- Summarizing feedback, notes, and transcripts
Yellow light, a human leads and AI assists
- Copy: you write the first draft, AI tightens it
- Any stat or claim: verify it yourself before it ships
- Imagery of real people: clear the rights first
- Voice critical lines like taglines: a human writes the final
Red light, do not
- Put confidential, customer, or unreleased data into public tools
- Ship anything without a human edit and sign off
- Generate the likeness of a real person without rights
- Pass AI output off as verified research or original citation
Before anything ships, five gates
The playbook in action
A concept scenario. "Fernway" is a fictional company I built to show the playbook working end to end. The thinking and the guidelines above are my real operating practice. The numbers below are the targets I would set and how I would hold the work to them.
The situation
Fernway is a mid market B2B SaaS company with a lean in house creative team of eight. A major product launch lands in 90 days, and leadership wants launch content across five channels, web, email, social, sales enablement, and paid, at roughly double the team's usual output. No new headcount. This is the 2026 squeeze in one sentence: more surfaces, same team, and AI sitting there as either a lifeline or a liability depending on who is directing it.
How I would run it
Before a single asset gets made, I set the system. I write the launch brief and make it the source of truth. I get all eight people onto one shared assistant loaded with the Fernway brand, so nobody is inventing the voice on their own. I draw the line on where AI runs and where a person starts, and I sit down with each report to build an AI plan for their role. Then the work flows the way it should: the ideas come from the team, and AI polishes, checks, and scales them out across every channel. Humans lead. AI amplifies.
What success looks like, the targets I would set
- The team's own ideas, produced further and faster, because their hours go to thinking and craft instead of grinding out production from zero.
- All five channels shipped for launch on the existing headcount, the capacity problem solved without lowering the bar or burning anyone out.
- A much shorter path from brief to first draft, so more of the calendar goes to iteration.
- Zero off brand or non compliant assets shipped. That is the number I would actually protect. Volume only counts as a win if the brand still sounds like itself and nothing crossed a legal line.
The point was never that AI made Fernway faster. It is that a directed team let eight people punch well above their size while the brand got sharper, not blurrier. That is the difference between a team using AI and a team being led through it.
About the author
Alisha Arnold is a creative director and marketing leader with 21 years of experience building brands and leading creative teams across B2B and B2C, including two full rebrands and global creative teams spanning time zones. Her campaign work includes a program that drove $250K in new revenue on a $52K budget and a live event concept that generated 1,200 leads at one of the industry's most competitive trade shows.
She writes about creative leadership, brand systems, and leading teams in the age of AI at thecreativedirectorlady.com.
© 2026 Alisha Arnold · thecreativedirectorlady.com · "Fernway" is a fictional illustration.