How I’m Using AI in My Video Production Business

I’ve been working in video production for 27 years, and I’ve never seen this much change happen so quickly. I’ve worked through the move from tape-to-tape editing to non-linear editing, the arrival of Final Cut Pro, the growth of the web and the rise of social media. All of those changed how we worked. What’s happening with AI feels bigger, and the pace is hard to keep up with.

Recently, I watched ChatGPT open Premiere, move into an embedded After Effects project and make a client’s changes. It then opened Illustrator to create something it needed, brought it back into After Effects and updated the project in Premiere. I just sat there watching.

That’s one example, but my main interest is how AI can help me run the business. I’ve been using it for websites, proposals, production planning and keeping track of what needs my attention. I’ve also been helping my coaching clients put it to work in their own production companies.

There are still frustrations, and there are things I’ve decided to keep doing myself. What follows is a look at what’s working for me, where I’m drawing those boundaries and how you might start exploring it in your own business.

Decide how you want to approach AI

AI brings up a lot of strong feelings. Some people are enormously excited about it. Others are worried about privacy, employment, creativity and what we might lose by handing more of our lives over to it. Those are serious questions, and I think it’s important to spend time working out where you stand.

When I’m looking at AI in a business context, I try to assess it as a tool. I want to understand what it can help me do, where it falls short and whether the result is useful. I explore the broader philosophical and ethical questions in my own time, while giving myself room to test practical applications in the business.

You’ll need to work out your own approach. For me, that has made it easier to experiment without turning every task into a debate about the entire future of the technology. I can try something, look closely at what happened and decide whether it belongs in the way I work.

The change has been exciting and unsettling

I was an early enthusiast, but I also found plenty to be frustrated by. ChatGPT could make things up, sound confident when it was wrong and praise my ideas in a way that felt excessive. I built useful workflows, while remaining very aware of the limitations. At one point, I came close to switching to another tool.

Then I tried a newer version that could take control of the computer and carry out tasks for me. I spent a couple of hours testing it that night, watching it work through things I’d asked it to do. It felt extraordinary. Tasks that had previously required me to explain every step were suddenly happening in front of me.

Alongside the excitement, I felt a sense of grief that surprised me. It was a sudden awareness that the world I was used to was changing, and that some familiar ways of working might be coming to an end. I went to bed feeling that, along with some uncertainty about what would come next.

The following morning, I was ready to get started. I had website changes and other projects that had been sitting in the too-hard basket for a long time. Now I could work through them. That experience became part of what I talked about in Monk Mode: How I Got a Year’s Worth of Work Done in a Month. Giving the work proper attention, with better tools available, helped me make progress on things I’d been putting off.

Using an AI agent as a virtual assistant

One of the things I’ve been experimenting with is ChatGPT’s Dots, using an AI agent as a virtual assistant. In my setup, it has access to information from the apps I use to run the business, including email and my calendar. That gives it context about my commitments, sales conversations and current projects.

I’ve been using it to give me a daily briefing: what I’ve committed to, where different sales conversations are up to, what’s happening in post-production and what my team needs to know. Having that information brought together saves me from checking several places just to get a picture of the day.

I’m still learning how to use it well. There’s a difference between asking for help with one task and setting up an assistant that can keep working through an agreed process. You need to be clear about what you want it to monitor, what information it can use and when something needs your attention. Even at this early stage, it’s been a substantial change in how I organise my work.

Better websites start with a proper design process

Web design has been one of the most useful applications for me. I’ve used AI to update pages, create case studies and develop landing pages. It has made changes possible that previously felt too time-consuming to organise.

But the first page I asked it to design was average. I then asked how a professional web designer would approach the project. It described a process involving the audience, an initial concept, the style guide and a mood board. I asked it to take me through that process, and the quality improved considerably.

That taught me something useful. The way you approach the work still matters. Give it a clear audience, a purpose for the page, relevant examples and the information a potential client needs. Review the design and copy, and keep refining them. A vague request gives it too much room to produce something generic.

Some of my coaching clients have used this approach to create landing pages for a particular industry or type of client. They take my landing-page framework, add their information and work through the process. It lets them test an idea and get a campaign moving much more quickly than they could before.

Speed is useful when it helps you act on a clear business decision. Before creating another page or offer, think about which services belong in your video business and who you want to serve. Then build a page that speaks to those people.

Turning scattered project information into useful documents

Production planning often involves gathering information from several places. There’s a client meeting, the brief, your own notes and a long email thread containing details you know you’ll need. Finding and organising all of that can take a surprising amount of time.

I’ve been using AI to bring that material together and prepare planning documents, including call sheets, shot lists and storyboards. It can get much of the initial work done, leaving me to read through it, fill the gaps and make the decisions that depend on understanding the project.

I think of it as getting me roughly 90 per cent of the way there. The final part still needs my judgement. Does the schedule make sense? Have we understood what the client wants? Is anything missing? A document can look complete and still contain assumptions that need to be checked.

Scriptwriting has been useful too, particularly for functional content such as training videos. With the client’s material and relevant research available, AI can help organise the information and develop a draft. I still need to make sure it says what we need it to say and will work for the people watching it.

Build a repeatable process for proposals

Proposals are another good example. I can bring together a meeting transcript, the client’s brief and my notes, then use them to develop a proposal. The improvement came when I gave the AI a defined process to follow.

I’ve created a proposal-writing skill, which is essentially a set of instructions for how I want that task done. It follows the structure I use, works from the information provided and helps turn the conversation into a clear production scope. I review it before it goes to the client.

If you regularly explain the same task to AI, it’s worth writing down your process. Show it an example of a good result, describe the steps and be clear about what needs checking. Then improve those instructions as you use them. You’re putting your experience into a form that can be applied again.

AI editing is becoming useful, but it still needs an editor’s judgement

I’ve tried some AI-generated video, but it’s not my main area of interest. What has caught my attention is the ability to work with existing footage, transcripts and editing projects.

The motion-graphics changes I watched it make involved text, images, colours and the order of the content. It moved between Premiere, After Effects and Illustrator and completed the changes correctly. A good editor could have done that particular job faster. Some of the delays were quite mechanical, such as struggling to click a control or needing another way to make a change. But other parts happened remarkably quickly.

I’ve also built a detailed process for creating interview paper edits. AI works from the transcript to help shape the story, then builds the edit in Premiere. That has been useful for case-study videos, where there can be a lot of interview material to work through.

One of the most interesting improvements came after I explained why a few selections didn’t work in the edit. The words looked good on paper, but the delivery sounded awkward when the sections were cut together. The process was then expanded to include an audio review, so those problems could be considered before settling on the selections.

That’s a reminder of why the transcript is only part of the material. The way someone speaks, their emphasis and the rhythm of a cut all affect whether the story works. I still need to watch, listen and respond to what’s there.

The things I’m choosing to write myself

I’m finding AI useful for email, but I generally prefer to write the message myself. I know what I want to say, how I feel about the situation and what the relationship calls for. Once I’ve written it, I might ask for help making the tone more friendly, helpful or respectful.

That lets me write quickly and then refine the message. It suits me better than handing over the whole conversation and hoping the result sounds like me.

Social media is another area where I’m keeping my own voice. When I recognise the familiar rhythm of an AI-written post or video script, I tend to lose interest. The wording might be polished, but I often feel I’m learning less about the person behind it.

Your posts can show people what you think, what matters to you and what it might be like to work with you. Those are valuable things to share. I’ve seen creative, thoughtful people with distinctive opinions default to generic scripts, and it feels like a missed opportunity. For now, I’d rather write those posts myself.

The result needs to be better

I’ve set a rule for my use of AI: the result needs to be better than what we would have created without it. It’s easy to focus on how quickly something was produced and overlook whether it’s actually helpful.

We’re receiving more briefs, schedules and interview notes that appear to have been generated with very little thought or review. Sometimes they create extra work because someone has to untangle the information, identify the gaps and go back to the client for clarification. A finished-looking document hasn’t necessarily solved the problem.

When we use AI on an edit or a planning task, I still want us to wrestle with the content. We look at what it has done, respond, ask questions and give feedback. That involvement is part of getting a good result.

I’ve also found it helpful for reviewing a long project contract and checking accounting work. In areas like these, I use it to identify questions and things that deserve closer attention. As with the rest of the work, I need to understand the findings and know when professional advice is required.

Start with one useful task and spend time learning

I think of learning to use AI a little like learning a language. You get better by spending time with it, trying things and learning from what happens. Reading about the tools can give you ideas, but working with them teaches you where they’re useful and where they still need help.

Choose one task in your production company that you do regularly or keep putting off. It might be preparing a proposal, gathering information for a shoot or improving a landing page. Give the AI the context, show it what a good result looks like and work through the task together. Review the output closely, then keep the instructions that helped.

You don’t need to rebuild the whole business at once. Start with something useful, develop confidence and let that experience guide the next step. Some of the coaching clients making good progress with this wouldn’t describe themselves as particularly confident with technology. Following a clear process has helped them get started.

The tools will continue to change, and so will the role you want them to play. Keep making conscious decisions about where they help, what needs your attention and which parts of the work you want to keep doing yourself.

Want help putting AI to work in your production company?

An increasing part of my work as a coach is sharing the processes I’m developing and helping production company owners apply them to their own businesses. If you’d like help with landing pages, proposals, production planning or repeatable AI workflows, you’re welcome to find out about coaching for video production business owners.