Inside the Pod: A Guide to Launching a Successful One-person Company

Thank you to everyone who is watching or listening to my podcast, How Do You Use ChatGPT? If you want to see a collection of all of the prompts and responses in one place, Every contributor Rhea Purohit is breaking them down for you to replicate. Let us know in the comments if you find these guides useful. —Dan Shipper

Was this newsletter forwarded to you? Sign up to get it in your inbox.

---

In the first week of April, I came face-to-face with the part of being a freelancer I hate the most: tax season.

The idea of wading through invoices, account statements, and 1099 forms fills me with existential dread that rivals walking into the high school cafeteria dressed head-to-toe in neon spandex.

I don’t like doing taxes and, for reasons both rational and irrational, it takes up an unnecessary amount of my time—time that I would rather spend writing.

I imagine that a founder running a bootstrapped startup must come across several tasks just like this—boring, yawn-worthy, and yet frustratingly unavoidable tasks that they must do if they want to build things, and earn money doing it. In this episode of How To Use ChatGPT, Dan Shipper interviewed Ben Tossell about how he uses AI to do the grunt work more efficiently.

Tossell is an experienced founder whose no-code platform Makerpad was acquired by Zapier in 2021. He currently writes Ben’s Bites, a daily AI newsletter with more than 100,000 subscribers, and invests in promising early-stage AI startups. Tossell has a knack for building one-person internet businesses that end up being sneakily big. 

In this interview, he told Dan how he’s incorporated AI into his workflow in order to:

• find great business opportunities,
• run a company efficiently, and
• evaluate the performance of a venture.

I think anyone interested in making money on the internet—as a freelancer, indie hacker, or solopreneur—will find their conversation useful. This essay dives into the nitty-gritty of how Tossell leverages AI to start, run, and evaluate businesses, complete with screenshots of his interactions with various tools.

How to find a great business opportunity with AI

Tossell uses ChatGPT as an expert strategist to analyze business models that he finds interesting. He believes that by analyzing the process, starting his own company in the same industry seems more achievable and less intimidating.

Recently, Tossell took interest in companies that curate and sell valuable data. He thinks this is the kind of business that has the potential to earn $50,000 in annual revenue with a very small team. Wanting to understand the industry better, he decided to study a leader: the market intelligence company Yardstiq, which offers access to exclusive interviews conducted by analysts with Fortune 500 companies that use software tools. Yardstiq’s sell is that the interviews will help prospective purchasers gain a deeper understanding of these tools before making big investments in them.

Tossell found a screenshot shared by the Yardstiq CEO on Twitter that captured a transcribed snippet of an interview conducted by its analysts. After uploading this image to ChatGPT, Tossel prompted it as follows.

Tossell: Make a reasonable guess as to what questions would be asked of the customer for each of the transcript sections. You can see one question on the transcript in bold.

Here’s a slice of what ChatGPT generated:

All screenshots courtesy of How Do You Use ChatGPT and Figma/Every illustrations.

Tossell wanted to create a blueprint list of questions that he could use as a starting point if he were to launch a similar kind of data company, a minimum viable product of sorts. To refine the questions that ChatGPT generated, Tossell uploaded a screenshot of Yardstiq’s landing page with customer testimonials to ChatGPT

Tossell: Now look at the quotes on this image. They are snippets of other responses from different customers talking about different software. Tweak the questions you just wrote based on what is mentioned in the image.

Here’s part of what ChatGPT came back with:

Tossell hoped that ChatGPT would examine the excerpted customer quotes, infer what questions could have prompted those responses, and frame similar questions for his own potential venture. However, he was disappointed because ChatGPT’s output was less detailed than its initial output. Tossell collected more context for ChatGPT to analyze, including more screenshots of customer testimonials, and prompted it further.

Tossell: Here are more images about Yardstiq, update the questions based off the information provided in these images.

Here’s part of what ChatGPT said:

Tossell observed that ChatGPT consistently generated similar questions across multiple attempts of using different resources to prompt it, leading him to believe that these questions likely resemble those actually used by Yardstiq while engaging with enterprise software buyers.

Dan notes that ChatGPT usually performs better when complex tasks are broken into separate steps. In this case, he recommends first asking ChatGPT to generate questions based on the new screenshots, and then prompting it to refine its previous question list using the newly generated questions. Dan and Ben decide to experiment with this approach live on the show. They upload screenshots for ChatGPT to analyze and then prompt it as follows:

Ben and Dan: In the screenshots I'm attaching there are snippets of various transcripts with customers of software tools. Sometimes the transcript includes a question that the customer is answering, it's in bold where included. based on the question-answer pairs that you can see for any transcript snippet that doesn't have a corresponding question, please try and generate a question that produced that answer. Be specific. Be thoughtful. Be detailed. I’ll tip you $2,000 if you do a good job.

Here’s a part of what ChatGPT generates:

Either the $2,000 bribe was effective or Dan was right about ChatGPT yielding better results when intricate tasks are divided into smaller parts. Dan and Ben are pleased that ChatGPT came up with new questions!

Dan is curious about what Tossell’s next step was after curating these questions, so they dive back into his chat history. Letting his curiosity lead him, Tossell embarked on a few side quests. First, impressed by Yardstiq’s landing page copy, he prompted ChatGPT to improve it.

ChatGPT:

Tossell also prompted ChatGPT to organize the data from the customer testimonial screenshots he had uploaded previously into categories that represent the different sections of an interview that a Yardstiq analyst would conduct.

Tossell: Based on the below titles, group the quotes and text from all the previous images into where you think they best fit.

• Introductions
• Evaluation Structure
• Solution Specifics
• Sales Experience
• Deployment
• Competitors
• Pricing and Packaging
• Post-Deployment

Here’s part of what ChatGPT generated:

Tossell spent some time prompting ChatGPT to be more specific before moving on to his next task—helping him build an unfiltered review site for AI products that evaluates them based on their features, pricing, integration, and deployment.

When Tossell asked ChatGPT for guidance on launching a business like this, ChatGPT generated advice around identifying a niche, researching competitors, etc.—in other words, Startups 101. Tossell, as an experienced founder, found this fairly basic, so he prompted ChatGPT to focus on the more challenging question of content collection for the review site.

Here’s a slice of ChatGPT’s response:

After remarking that he had already implemented some of ChatGPT’s suggestions, Tossell notes that this response sent him down another rabbit hole about a review site with a potentially similar business model—Glassdoor, where employees leave anonymous reviews about their workplace with the aim of assisting job seekers. Tossell started another chat to prompt ChatGPT to find out more about the platform and its practices. His first question was about Glassdoor’s founding.

ChatGPT:

Tossell: How did they/do they monetize?

ChatGPT: 

Tossell thought it was interesting that one of Glassdoor’s revenue streams was charging employers to enhance their brand on the site because it seemed to conflict with the platform's objective of giving job seekers unfiltered reviews about a workplace. He prompted ChatGPT to understand how Glassdoor’s business model has evolved.

Tossell: How did Glassdoor get reviews on the site initially? Did they run any incentives?

Using AI to run your company efficiently

After sharing how he uses AI to find great opportunities, Tossell shares how he uses it to run his business, Ben’s Bites, especially the new tier, Ben’s 2.0, that he launched a couple of months ago. Part of Tossell’s premium offering is longform pieces based on interviews that he conducts with founders on how they’re using AI tools.

Tossell walks us through his process of transforming an interview transcript into a useful article. First, he downloaded the transcript from note-taking app Supernormal and uploaded it to one of Google’s Gemini models, prompting the model to help him write a blog post.

Tossell: Find my document titled “Eric Vaughan/Ben Tossell transcript” and highlight key topics that I should write about in a blog post. I write posts about how businesses are using Al internally and in their products. Include quotes in each section of the blog post draft.

Here’s part of what Gemini generated:

Tossell believes that Gemini has provided a solid foundation for him to begin drafting his post. To gather additional insights, he decided to input the same transcript into other AI models, such as GPT-4 and Claude. The next step of Tossell’s writing process was to copy and paste all these AI-generated drafts into Lex, the AI word processor Every incubated.

Tossell took advantage of a feature in Lex which enabled users to create multiple versions of the same draft. He combined all the drafts produced by the AI models in a single version, making it easier for him to review the consolidated material.

Version 1 of Tossell’s draft that consolidates the AI-generated insights:

Tossell created a second version of the draft in Lex where he started drafting the blog piece.

Version 2 of Tossell’s draft:

Source: Lex/Every illustration.

Once Tossell completed the initial draft of his blog post, he leveraged an AI-powered tool within Lex to assess the quality of his writing. This feature analyzed the draft and provided feedback focusing on two key aspects: brevity and readability—and just like that, the piece was ready for Tossell to publish.

Source: Lex/Every illustration.

Using AI to evaluate a business

In the last segment of the interview, Tossell shares how he uses AI tools to analyze business data.

Tossell: I surveyed subscribers of my newsletter and asked them why they continue to stay subscribed. Their answers are in the CSV file attached. Focusing only on the information in column A, can you read the answers, then summarize how the majority of people feel? You can ignore any outliers or anomalies.

ChatGPT:

While noting that ChatGPT did not follow the instruction of discounting outliers in the data, Tossell prompted it to go even deeper.

Tossell: Run a sentiment analysis on the data in that column.

Building a career on the internet can be daunting, and this interview reminded me how AI can make the process easier. In fact, I’m almost looking forward to tax season next year—so long as I can figure out how to use artificial intelligence to make it less painful!

---

Rhea Purohit is a contributing writer for Every focused on research-driven storytelling in tech. You can follow her on X at @RheaPurohit1 and on LinkedIn, and Every on X at @every and on LinkedIn.