When AI Does the Work, Someone Still Has to Think
*Ownership Essentials with Maya Alexander, Founder of State of Sonder*
The pitch is simple enough. Subscribe to a few tools, feed them your brand info, and walk away with a week’s worth of content by Tuesday afternoon. No agency retainer. No shoot days. No waiting on a freelancer who’s juggling six other clients. Just output, fast, on demand.
Most business owners who try this discover the same thing around sixty days in. The output is there. The results are not. The posts go out. Engagement stays flat. The videos look a little off, somehow. And the owner is now spending weekend hours reviewing AI drafts instead of the weekend hours they used to spend on a vendor invoice or a staffing problem. The time didn’t disappear. It moved.
In this Ownership Essentials conversation, I sat down with Maya Alexander, founder of State of Sonder, an AI media and content production agency serving small business and direct-to-consumer brands. Maya works at the intersection of paid advertising, AI-generated creative, and brand storytelling. She spends her days helping owners figure out what AI can actually do for them versus what they only think it can do.
The Gap Between Output and Outcome
Maya started State of Sonder because she saw two things happening at the same time. AI was removing the cost barriers that had kept high-quality content production out of reach for most small businesses. And business owners were walking into that opportunity with expectations that didn’t match reality.
The expectation, stated plainly: one subscription replaces an entire creative process. Immediate publish-ready work. Brand consistency without having to explain the brand. Fewer outside professionals. Lower costs, full stop.
The reality she runs into: owners assume that speed of generation equals speed of final approval. You can produce output quickly. That output can still be generic, inconsistent, and strategically off. An AI model can generate something technically impressive that has nothing to do with what the business actually needs to say. If no one with judgment is in the room, that content goes out anyway.
Sure, AI gives you acceleration. But acceleration without direction just gets you to the wrong destination faster.
What “AI Slop” Actually Costs
We spent time in the conversation putting a clear definition around AI slop, because it’s one of those terms people use to mean different things. For Maya, it’s the six-fingered hand, the character whose face changes between frames, the product image where the logo is distorted and the proportions are wrong. Generic language, inconsistent visuals, anything that signals to the audience that no one reviewed this before it went out.
Does it actually hurt, or is it just aesthetically uncomfortable?
Her answer: it depends on the industry, but for product brands, yes, it damages engagement. Trust is earned slowly and lost quickly. Audiences are developing a faster reflex for detecting content that wasn’t made with care. LinkedIn has a button for it now. That’s not an accident.
Beyond the aesthetic problem, there are harder risks that most business owners don’t think about until they’re in trouble. False advertising claims are easy to generate and easy to miss. AI models hallucinate facts, statistics, and quotes, and if an owner trusts the output without reviewing it, those hallucinations go to print. In my own practice, I’ve watched compliance-trained AI tools insert financial performance claims into franchise marketing copy (claims that violate FTC regulations) after being explicitly told not to. Then the AI apologizes and does it again in the next draft. The cleanup responsibility lands on the owner every time.
Maya also raised something that’s coming but hasn’t fully arrived yet: likeness and consent issues around AI-generated models. It is currently possible to generate a digital model that resembles a real person without that person’s knowledge or agreement. The regulations around this are, in her words, quite a few steps behind where they need to be.
The Human in the Loop Is Not Optional
The phrase Maya returns to throughout the conversation is “human in the loop,” and she’s specific about what that means in practice. It’s not simply having someone glance at content before it publishes. It’s a defined role with defined responsibilities: setting the objective, supplying accurate brand and product information, establishing visual and verbal boundaries, selecting outputs, directing revisions, rejecting work that is technically capable but strategically wrong, and fact-checking before anything goes out.
Someone in the business needs to own AI orchestration. As AI agents become more capable, they also become more like employees: they can break, they can hallucinate, they require active management and guardrails. The idea that AI removes the need for human oversight gets it backward. It changes the nature of the oversight required, but it doesn’t reduce it.
Her practical advice for owners trying to get this right: stop adding tools. Most of the AI content platforms running right now are wrappers on the same underlying models. Adding a fifth tool doesn’t add capability; it adds coordination overhead and more places for things to break.
What the Right Partner Looks Like
I asked Maya what a business owner should look for when trying to find someone to serve as that human in the loop without paying agency retainer prices.
Her first filter: if they start by talking about what they can do before they’ve asked about your business, keep looking. A strategy that isn’t built around your specific situation, your audience, your story, your compliance requirements, is just someone else’s template applied to your name. That’s a faster route to AI slop.
She also made the case for spending real money upfront in places that feel counterintuitive. A traditional photo shoot of your physical location, your products, your actual team gives an AI-assisted workflow the reference material it needs to produce consistent, accurate visual content over time. Skipping that foundation doesn’t save money. It means the AI has nothing accurate to work from, and you end up correcting that problem one bad output at a time.
What the Balance Actually Looks Like
When this works, and Maya is clear that it can work, the owner is no longer choosing between disappearing from their content channels or becoming a full-time content creator. There’s a repeatable system: a defined workflow, a clear brand voice the partner understands, a schedule that delivers without constant owner involvement, and enough creative range to test what’s resonating with an audience.
The owner gets back to the work they’re actually good at. The content keeps moving. The brand stays consistent. The AI does what it’s actually built for: removing the blank-page problem, accelerating ideation, generating first drafts that a human then shapes into something worth publishing.
Maya’s closing thought: AI can’t replace the story behind a business. It doesn’t know why the founder took the leap, what they gave up to build the thing, or what makes their version of this different from the ten other businesses in the same category. That’s still the owner’s job to know and to tell. AI can help carry the story once someone figures out what it is.
If you’re an owner trying to figure out where your AI content strategy is actually breaking down, or whether you even have a strategy rather than just a subscription, Maya and her team at State of Sonder are worth a conversation. You can reach her at stateofsonder.com.