The "Don't Be Cringe" AI Strategy Fueling a 6-Figure Solo Business
Or How To Spend Less Time Managing People And More Time Managing Content
It’s nice to see your ideas validated, which is what happened to me this week.
I came across an interview with Barbara Jovanovic on Youtube that showcased the massive opportunity that is AI for solopreneurs.
Barbara runs a successful six-figure marketing agency called startup cookie, which she runs pretty much by herself these days.
How? By leveraging AI and automation to do most the work that she used to need to outsource to roughly 10 different freelancers.
Not only is AI a lot cheaper, it’s also a lot quicker and more fun to flesh out your ideas and generate first drafts in real-time working with AI than it is to have to manage lots of humans and wait for weeks for them to get back to you with something that isn’t exactly what you had in mind anyway.
The column on the left contains all the people she had to manage before switching to an AI-first workflow in which she can run the same content for a fraction of the costs:
Barbara is a perfect example of someone who’s mastered AI by having a clear vision.
She had a marketing process that was working but in which she was spending most of her time managing freelancers and waiting for them to get back to her with what was often mediocre content.
So her decision to replace these freelancers with AI tools was a no-brainer.
It made her work more fun, and increased the speed and quality of client results.
To be clear, process + AI isn’t always a winning formula.
I think that a key part of her success comes from her leveraging AI to its strengths.
Instead of using AI tools and automations to generate bland and generic content, the starting point for her client work is always the pure unfiltered thoughts of the tech founders and CEOs she works with in her agency.
She uses podcasts in which her clients have featured, and supplements those with monthly one-hour interviews with her clients to get the seeds for the marketing content her agency produces for her clients.
By staying as true as possible to the real, lived experience of the people running these startups she’s able to keep content generated with the help of AI fresh and original.
AI is used to quickly and craftfully repurpose these initial seeds across the myriad different marketing channels and platforms a company needs to maintain these days.
Another key part her success is that she also understands how NOT to use AI.
Take a look at one of her prompts for repurposing existing content (bold is mine):
Rewrite the following text in a more punchy, to the point way. It should match the tone of voice and writing style of a tech founder. But NOT CRINGE and not cliche.
Do not use metaphors, catch phrases or jargon. Do not use the words "ensure", "in the realm of", "in the world of", "Remember,". Make sure you don't lose any important insights or information from my original draft. Keep the text valuable and insightful. Only use the information provided in my original draft. Edit it for spelling mistakes, grammar issues, punctuation, and generally for readability and flow. Make your best effort.
Important: Never add a title or bold section; LinkedIn doesn't support formatting
Important: Don't add emojis, or in very limited cases (e.g. 1 / 2 /3 emojis) to structure ideas
Important: The first sentence is called the hook and that should be catchy so people start to read the text
Important: Always add spacing between sentences so it’s easy to read.
Text to Rewrite:—Barbara Jovanovic, 20 Prompts to Scale Your Marketing
This one prompt contains 14 instructions to NOT do something—in fact, it contains way more negative than positive instructions! (I particularly love the “don’t be cringe” guideline, we could all use one of these! :))
Just wanted to share this here as a great example of someone who has successfully used AI to leverage her business expertise. She built a six figure marketing agency that would have—until very recently—required an entire 10 person team.
Check out the full interview with Barbara Jovanovic here on Youtube..
PS If you want to leverage AI to build your own six-figure business but don’t have the time to spend countless hours experimenting with AI to find what works for you—send me a message!
This week in AI
It’s been a busy week in terms of AI news and releases—as agentic workflows are slowly creeping into all consumer AI products, “agentic” is becoming a strong contender for the word of the year:
The 2025 Stanford AI Index highlighted accelerating AI adoption (78% of businesses) fueled by record US private investment ($109B) and confirmed productivity boosts across the workforce. AI performance continues to surge on demanding benchmarks, even outperforming humans in some timed tasks, while becoming increasingly embedded in daily life from healthcare to transport. Concurrently, AI is becoming significantly more efficient and accessible, driven by falling costs, improved energy efficiency, and increasingly capable open-weight models closing the gap with proprietary systems.
Anthropic has enhanced Claude with a new 'Research' capability and Google Workspace integration, aiming to make it a more powerful collaborative partner. The 'Research' feature enables Claude to agentically search across both internal documents and the web, automatically exploring different angles of a query with citations. These updates, currently rolling out in beta for paid users in select regions, help automate information gathering and analysis, saving significant time.
OpenAI has released its new GPT-4.1 family (including mini and nano variants) via API, boasting superior performance, especially in coding and instruction following, compared to GPT-4o. These models feature significantly improved capabilities like enhanced long-context comprehension up to 1 million tokens, better code diff generation, and more reliable format following. Available only through the API with a knowledge cutoff of June 2024, GPT-4.1 replaces the GPT-4.5 Preview, which will be deprecated in July 2025.
OpenAI released its latest reasoning models,
o3ando4-mini, described as their smartest models to date, trained to think for longer before responding. The powerfulo3model pushes the frontier in complex coding, math, science, and visual tasks, while the smallero4-miniis optimized for speed and cost-efficiency, particularly in math, coding, and visual analysis. Both models represent a significant advancement by agentically using tools like web search, data analysis, and image generation directly within ChatGPT.Google has made its advanced video generation model, Veo 2, generally available for developers via the Gemini API and Google AI Studio, enabling the creation of 8-second, 720p videos from text or image prompts with enhanced realism and cinematic control. Gemini Advanced subscribers can now also generate videos directly within Gemini using text prompts. Additionally, the Veo 2 model powers the new Whisk Animate feature in Google Labs, allowing subscribers to turn images into short video clips.