Agents Took Over My Partner's Job. Nobody Took Over Mine.

For about fifteen years I ran GlockSoft with a partner. Our product was GlockApps, an email deliverability tool. He wrote the code. I did everything else: marketing, support, answering customers. There were no agents then. In 2024 life circumstances made me sell my half.

Now I'm building ViralFaceless.io(opens in new tab), for people who run faceless video channels, and ClipWright.io(opens in new tab), where one agent command turns a script into a vertical lip-sync video. Dimantika, the company behind them, has two co-founders: me and my wife, who handles the accounting. On the engineering side I work solo, if you don't count the agents. They write most of the code, which comes with its own problems. I've written about agents that report "done" when they aren't and about the rework that eats the speed they give you.

On a normal day I have three or four agent sessions running at once. When the work moves, that's what keeps me going. I want to make video tools that a creator or a marketer can use without being technical, and I'd like to make my own product ads as a video series too. I also help my wife run her audio-stories channel, with agents there as well. They make the work several times faster, as long as the person driving them knows the field.

The part my partner used to do, the code, is the part that got automated. The part I led, marketing, is still on me. My partner never argued marketing with me; it wasn't his field. Now that everything else moves several times faster, I notice that gap more.

Andrew Ng describes the bigger version of this. In an interview with Marina Mogilko(opens in new tab) in August he said "the cost of building has plummeted and so the challenge is shifting to deciding what to build." He calls it the product management bottleneck. My partner's side of the work got cheap. Mine is now the bottleneck.

What an argument would be for

Positioning, channels, copy. Those are the three places where I want someone to push back.

Positioning is the question of who the product is for and what it replaces in their life. It is hard to answer alone, because you always agree with yourself. You wrote the sentence, so it sounds right to you.

Channels come down to where the next month goes. From inside your own head every channel looks promising. You need someone who will say "that audience isn't there" and make you prove otherwise.

Copy comes down to whether a headline says anything at all. I can't judge my own. I know what I meant, so I read the meaning into the words whether it's there or not.

Code has tests. A positioning line has only the person across the table.

Why the agents don't fill that chair

Agents do a lot of what I used to need other people for. They'll draft landing page copy and give me headline variants if I ask.

What they can't do is argue marketing with me.

If I just ask whether a headline is good, I get something average back. A generic answer. It gets better when I give the agent examples that actually took off, from competitors and from people I follow, so it has something to measure against. Where agents really earn their place is analysis: taking apart what worked and why. That part I like.

Ng describes a related problem. As a brainstorming partner, he said, AI often comes up with "one or two good ideas, two or three mediocre ones" and "four atrocious ones", and it doesn't know which ones are awful. Someone with context does. Ng calls it "a massive context advantage": you saw the customer's face when something didn't land.

An argument needs someone with a stake in the answer who is willing to be annoying about it. Someone who remembers what we decided last month, says so, and tells me no before the market does.

Right now nobody does that part for me.

Where it showed up

Tibo posted about his exit this week, and I left a reply under it. Within eleven minutes another founder had joined the conversation with me.

So I asked the same question on my own account: if you went solo after years with a co-founder, what replaced that person?

What I'm doing about it

There hasn't been much marketing yet. The next step is listing both products in startup directories. What I already do is sit down with agents and take apart what took off for other people.

In the same interview Ng said "All of my marketers know how to code," and that he asks marketing candidates what they've built. By that measure I'm already the new kind of marketer, with two products in production to show for it. It still doesn't give me anyone to argue with.

Hiring a marketer is on the table, but after sales start. Until then it's me and the agents. Even that has given me a few ideas, and it has made one thing plain: without marketing there are no customers, and without customers there is nobody to tell you what's wrong with the product.

So, a question

If you build without a co-founder on the marketing side, who argues with you?

A friend who runs their own company? A paid advisor? A founder group? Your first customers? Your own posts, and whoever bothers to reply?

I'd like to hear what actually works for people, not what sounds good in a thread. Tell me what you use, and whether it disagrees with you often enough.