How to start a ChatGPT Wrapper Company

Presented by

Hey — It’s Hussein 👋

I have to be honest. Getting into the mindset to share something useful on social and write this newsletter has been hard this week. I am not going to get into Middle East politics, but it’s difficult to ignore what’s going on over there and carry on with my AI and tech bubble.

My heart goes out to the families and children of both sides who want a normal life but might be in the middle of a deadly war for years to come. Let’s all hope it doesn’t go there. 🙏

Today’s topic is going to be particularly interesting to those wanting to build on top of ChatGPT. Let’s talk about ChatGPT Wrapper Companies.

Before we begin… a big thank you to:

I am purposely adding the word “company” to ChatGPT Wrapper because I believe there is a difference between ChatGPT Wrapper and ChatGPT Wrapper Company. Here’s how I see these:

ChatGPT Wrapper is a tool that helps make it easier to use the OpenAI API. Usually, these are open-source projects. They give an interface into the API, could have embeddings or fine-tuning built-in, etc… Made by developers for developers. AutoGPT is a great example of this.

ChatGPT Wrapper Company is a company that is building a tool that uses ChatGPT in the backend for a specific workflow or use case. I see these as having two very specific parts: they perfect the prompt for a specific use case, and they create some workflow.

How to build a ChatGPT wrapper company?

If you are thinking about doing this, you are most likely not alone! If you browse Product Hunt on any day, you can clearly see that ChatGPT Wrapper Companies is what most founders are focused on these days.

So, how do you separate from the pack, and what should you build?

1. Idea

Just like any startup, you have to solve a real problem and have a real use case. Is there any part of your job or day-to-day that can be improved by using ChatGPT? Do you find yourself using the same prompts over and over? Come up with a list of ideas and move to the next step.

2. Prompt

Use the OpenAI Playground and see if you can create a prompt that truly helps solve the problem or complete the task you are looking to build around. Read this intro to prompt engineering to come up with a good prompt that you could use.

3. Fine-tuning and Embeddings

I haven’t covered these topics in detail yet, but I will soon. If you take your use case, your problem, and your prompt(s), are there additional data sets that would be useful to get even better results? Can you start pulling this together so you can fine-tune the OpenAI API to use it in its response?

4. Simplify a Workflow

In addition to the above, is there a certain workflow, approval process, series of steps, etc., that can be automated or handled with a tool? I find that having this, in addition to a perfected prompt, is what can help set your product apart.

5. Build an MVP

Armed with the info from the previous steps, you can start building your product now. Here you have a few options:

1. If you are an engineer, good on you, start building.

2. If you are not an engineer but are technical enough, see if you can use ChatGPT to help you build it! This is the path I am taking for an experiment I am working on.

3. Coming soon: use a no-code tool that can help you build this. I have not come across anything that works great just yet, but I am certain the no-code tools are working on making this a reality very soon.

4. Hire a team to build it! I can provide referrals if you need help, but that is the last option for now.

6. Test, Launch, Learn, Iterate

Then, the fun begins: show your tool to others, have them use it, ask for feedback, iterate, and keep going. Don’t forget to let me know, too. I am happy to share it!

Example ChatGPT wrapper companies

Here are some examples of ChatGPT Wrapper Companies and the use cases they focus on.

VenturusAI - enter a business idea, get feedback, a SWOT analysis, market analysis, and other info you would typically find in a pitch deck, marketing plan, or business plan. Let’s break it down:

• Use case: analyze a business idea.

• Prompt strategy: first, they figured out what is important to their users, which are most likely founders thinking about ideas or investors analyzing startups. Then, they had to perfect each prompt to get the best possible results.

• Fine-Tuning and Embeddings: they may already have done this, but they are most likely also working on supplementing ChatGPT prompts with their own data set to help get even better results.

• Workflow: end-to-end process of getting feedback on an idea and creating a plan with the ability to iterate on it.

eduGPT - built for teachers to help them create lesson plans and come up with activities, quizzes, and tests. Let’s break it down:

• Use case: help teachers create a lesson plan.

• Prompt strategy: by grade and subject, they’ve most likely had to create different prompts in order to provide relevant information to the teachers.

• Fine-Tuning and Embeddings: sample lesson plans, sample quizzes, activities, etc., can all be used to improve the model.

• Workflow: break-down of lesson plans by grade and subject to walk teachers through creating them.

If you are working on a ChatGPT Wrapper Company, reply and let me know. I would love to learn more about what you are working on and share it with the GPT Hacks community.

See you next week — Hussein ✌️

P.S. If you’d like to sponsor, reply. 5.5k founders and entrepreneurs are waiting for you. 

P.P.S If you live in or visit San Diego, sign up here to be notified of tech and founder mixers I organize in the area. Would love to meet you there.