Co-creating with AI: why human vulnerability and AI editing win together

More human, less AI. Starting this week, this blog goes down to one post a week, from three. Content in an AI world needs stakes attached, because human-sounding writing is free now. The way I'm doing that is co-creation: AI does the research, helps organize my thinking and edits what I've said, and every thought in the post comes from me, including the parts I'd rather leave out.

Why is Van Dycke going from three blog posts a week to one?

For months this blog has published three posts a week. That was on purpose. When you're building a body of work from nothing, volume matters, and three a week got us to something like critical mass: enough on the page that a founder who found one post could find ten more.

Now I want to drill down into one topic at a time.

The honest reason is that volume pushed me to lean on AI more than I wanted. Three posts a week means more of the drafting and some of the research gets handed to the machine, and less of my own thinking gets into the words. I'll own that.

I also want to be fair to what we did. There hasn't been a single article on this site without an interview with me behind it. Every post has original material from how I run my consulting and stories from my own experience. I drove every one of them. I just haven't been happy with how much of me made it onto the page.

When I look back at those posts, four things are thin: depth, accountability, vulnerability, and stories of failure. You can probably get better information from AI than from my brain. You can't get those four from AI.

So starting this week, Van Dycke is going down to one blog post a week. One post, built from a longer interview, with something in it I could be held to.

What's the failure I never wrote about?

If I'm going to say failure stories matter, I should go first.

For a long time I carried a real insecurity about proving the return on the consulting we do. That's a fair thing for a client to ask about. A good business owner buying from a B2B service provider wants the investment to come back. If you're installing furnaces, the new furnace should be efficient enough to pay for itself over the old one. Obvious. When someone hires a consultant, there should be a tangible return too.

The trap I was in was believing that return had to be revenue. Our work is going to make our clients money. Eventually. But it takes a long time to show up, and I was selling it as if the revenue would be the proof.

It didn't always show up on the timeline people expected. I have ended at least two contracts well before they were finished because the client was unhappy with me. Both were fairly early in the engagement, and in both I did a bad job of managing expectations. Looking back, I'm not sure I could have managed them well, because the expectation itself was wrong. If you set the wrong expectation, there's nothing to manage.

What changed it was a conversation with one of my sales coaches, Mark Repkin. He described return on investment as having three different expressions:

Revenue. If you put money into advertising, it should show up in your revenue. If it doesn't, the advertising isn't working.
Risk management. Change, turnover, a market shift. These can derail a company, and they're very hard to see from the inside.
Time. Either getting more efficient, or getting the founder out of something.

The third one is where we live. Our goal with a client is to get the founder out of sales and marketing so they can focus on the work they actually want to do. Sales and marketing are the two growth engines of a business. Sales moves fairly fast. Marketing moves very slowly. You need both, and because they're so critical, they're the two seats founders find hardest to leave.

When I realized our return is time, it changed everything. I talk about the work differently now, and more confidently.

What I didn't do was write about it. It's hard to admit in public that we were focused on revenue and didn't hit revenue goals for some clients. It's harder to explain why you don't have great case studies, when the answer is that you were measuring the wrong thing. I kept that out of this blog because I thought it would cost me something.

In reality, we were looking in the wrong place.

These days I say it up front. Yes, we can build you a new website. Yes, a website needs to make you money. Depending on your industry you might see a fast return, but the odds are better that it's a long play. You'll know you have a great website and good content going out, and our team does really good work, but the revenue will take a while. Some people hear that and say no. They don't want to put that much money in up front, or the cash flow isn't there. That's completely legitimate. It's all about expectations, set at the beginning.

What does the research say people actually respond to?

Every week we run research on the content that's relevant to you, the reader. This week was no different.

That report is generated entirely by Claude. I've even taught other people how to build one. Its job is to give me access to what readers are dealing with right now. I don't want to write for the sake of writing, and I don't want to write only about what's important to me. I want to write about what's important to the people reading.

This week it turned up three things.

Confessions beat production. On a Reddit forum for agency owners, one owner admitted he'd finally fired a client worth about 4% of his gross who had been messaging "this is not working" on every bad day for months, and then wanted to negotiate the price down. That one embarrassing admission gave everyone else permission to tell their own versions. Another owner described a client that grew to around 30% of their gross. "He knew it. He was super pushy and treated us like internal staff." The thread was full of owners telling the truth about things that went wrong.

Polished volume gets silence. Some of the channels aimed at the trades are publishing videos that run 45 to 85 minutes. Our research found almost no comments on them, often none. That's a lot of content, and nobody is answering it.

Nobody's watching the human side. A very well-known marketing company posted a video about taglines. In the comments, owners were posting their own taglines and asking, one after another, "Does this make sense?" Nobody answered them. That's like turning on comments on your blog and never taking the time to reply.

We see this over and over, and people still take the shortest path to content. If it looks like AI can do it for you, great, ship it. And mostly, nobody responds.

Why are failure stories catching on?

There's a whole movement called Fuckup Nights (you'll have to excuse the language). Winnipeg has one now, which is my city, and I'm looking forward to attending my first. The format is simple: entrepreneurs and business owners get up on stage and talk about how they messed up, instead of their golden, shining moments.

At most networking events and professional events, people talk about their successes. And it's funny, because even when the gurus and the keynote speakers talk about their failures, it's usually to sell their book. Which is a success. That gives the whole thing a marketing quality that makes it a little suspicious.

When regular people stand up and say "yeah, we really messed this up, and it still hurts," that is gaining traction. We learn better from it. And it's human. Those stories need to come out.

Where does AI belong in your content?

I'm a big believer in AI. I'm always going to use it to help produce content. I'd like it to be more like co-creation than robot creation.

The best uses of AI in content are research, helping you organize your thoughts, and editing those thoughts after you've written or co-written them. In that order, and in that role. The thinking stays yours.

So if you go looking for the Claude watermark on this post, you'll find it. Claude interviewed me for it. Claude drafted it from what I said, and it'll edit the draft. What it didn't do is come up with any of the thoughts. Every one of them came out of my mouth, and I directed where they went.

Sales is still a pretty human business, especially with more expensive, high-ticket sales. Marketing is where it's easy to get lazy with what you produce. That's the part to watch. As AI makes you more efficient with systems, you still have to represent the heart and soul of your company, and that part has to be human.

If you want to see how I handle the AI question when it comes up with a client, I wrote about what to say when a client asks if you use AI.

Putting it to work this week

I don't want to write for the sake of writing. Every post should give you something you can put to work right away. This week it's this: get more human in your marketing.

Put AI where it's most effective. Look at the last few things you published and note what AI did on each one. Research, organizing your thoughts and editing are good jobs for it. If it also came up with the thinking, take that job back.
Get your own thinking out first. Before your next piece, talk it through out loud, or have AI interview you one question at a time and draft only from your answers. That's how this post was made.
Tell one failure. Pick something you got wrong and have avoided talking about. Say what happened, what it cost, and what you do differently now.
Answer the people who reply. If you allow comments, answer every one this week. A question sitting unanswered under your post tells people nobody's home.

Human interaction first. Human thinking first. Human vulnerability first. That will connect you with your audience far better than producing a vast amount of content.

What am I committing to?

From here on, every post on this blog carries one thing you can put to work right away, and every thought in it comes from me. AI does the research, the organizing and the editing.

We'll check back in January 2027 and report whether that held, including anything that didn't.

Sources

The links below were found by the AI research tools we use and checked before publishing.

Frequently asked

Should a small business use AI to write its blog?

Use it for research, for organizing your thoughts, and for editing what you've already said. Keep the thinking, the opinions and the stories yours. Readers can't tell much from how smooth a post sounds anymore, because AI made smooth free. They can tell whether there's a real person behind it with something at stake.

What does co-creating content with AI look like?

For this blog, AI runs a weekly research report on what our readers care about, then interviews me one question at a time. It drafts only from my answers and edits the draft. I decide the topic, the position and the stories. If a thought didn't come from me, it doesn't go in.

Why share failure stories in business content?

People learn better from what went wrong, and a real failure is something AI can't make up for you. Owners respond to it. In this week's research, the threads where owners admitted something embarrassing drew far more honest conversation than long, polished videos did.

What is the ROI of hiring a consultant?

It can show up three ways: revenue, risk management or time. For founder-led firms, our work mostly returns time, getting the founder out of sales and marketing. Revenue usually follows, but slowly. Agree on which kind of return you're buying before you start, because the wrong expectation can't be managed later.

Ready to look at the architecture honestly?

Book the conversation. We'll tell you what we see, and whether the work we do fits where you are.

Book the conversation →