How To Build An AI Startup In 2024

The gamut of options from single-owned to bootstrapped and VC funded

Skanda Vivek
4 min readJan 5, 2024


AI has had a boom in 2023 — from the early adoption of ChatGPT — to the emergence of highly competitive open-source models like Llama, Mistral, and the emergence of multimodal models like GPT4-V and Gemini. However, while the public has already started to take notice of AI, I believe 2024 is the year Generative AI startups — taking advantage of customers that want to adopt AI at lightning speed. No AI startup is the same due to the different types of use-cases and customers, but here I categorize the different types of startups by size and scale.

Consulting and AI Automations

I’ve seen cases of folks with a few years of technical and managerial experience in the industry decide to pursue consulting. The great thing about this is that you could start consulting as a side-hustle, and slowly grow it during off hours. When you have a critical mass you could switch to consulting full time if you wanted to. Consulting can also be very flexible. You work the hours you like, and most of the time work remotely.

However, in the flexibility there also lies issues. The single person consultancy can’t really be scaled and the only way you earn more is by putting in more hours — which in my opinion, reduces some of the attractiveness of starting a consultancy in the first place. Another downside of full-time consulting is the lack of exposure to building things at scale as part of a larger team. One of the main advantages of working in a large organization is that you are a part of building products at scale, and get to interact with like-minded, bright folks on a daily basis. By leaving your job, you might lose the opportunity to steer the incorporation of new AI features in larger organizations, so full-time consulting might mean that you lose your industry edge. However, consulting can also be a nice bridge to a more scalable AI startup idea as you now have the flexibility to pursue your own ideas.

Another type of opportunity is helping organizations create their own GenAI products — most popular being customer chatbots. Youtubers like Liam Ottley have popularized no-code solutions for building chatbots. Tools like, chatbase, and Stack AI have made deploying chatbots extremely simple. Due to the boom in smaller companies using AI and the variety of use-cases for chatbots to answer customer questions, this is a great space for startups.


You’ve probably heard of software-as-a-service (Saas), micro-Saas are built by individuals or smaller groups of individuals, and are built around customer niches. One example is a meme generator, Supermeme, turning text into memes. Another example is ShipFast, that helps with JS boilerplate for building a Saas. PhotoAI uses AI for photoshots from uploaded selfies.


A report by the A16Z team showed that out of the 50 top GenAI products, almost half are bootstrapped that haven’t raised any funding yet. This is particularly relevant right now as in the past, founders with ideas were able to raise funds without prototypes. But now, VCs are increasingly wary due to the economic climate. Another advantage is that by running a profitable, scalable business without relying on VC funding — you own more of the company and can get higher non-diluted returns as the startup valuation goes up, and can make key decisions with less scrutiny.

I’ve come across multiple Bootstrapped companies that aim to revolutionize industries using AI. EMAlpha is one example — that aims to revamp how financial trading is done across emerging markets using AI.

VC funded

Last but definitely not least, let’s talk about the type of startup that is built to scale right from the beginning. There are a lot of places to obtain funding. Currently, to obtain funding you need to show a prototype that has obtained some traction or initial customers, with the potential to grow. Once you have a prototype, you can hopefully get seed funding to grow the product a bit more over a year or so. The next step is Series A funding which lies in the few million dollars range. This would give you a few more years to hire full-time employees, and really build a product that is competitive.

Investors bank on the value of the company increasing in subsequent funding rounds and sell-offs/IPOs so that they get a good return on their investment — typically more risky than the stock market, but much more rewarding if you know what you are doing. Some good companies in this space include, for enriching data annotation using AI (in their website they claim the world’s best annotator). There are also interesting startups around evaluating, monitoring, and securing LLM outputs and include companies like Giskard and Uptrain. There is also a whole gamut of AI startups that aim to make LLMs more easy for developers to integrate into workflows (Langchain and LlamaIndex are popular examples).


Hopefully you’ve seen that there are several paths to AI startups. The ideas and companies I mentioned are just scratching the surface — so be bold and make something awesome!

If you like this post, follow EMAlpha — where we dive into the intersections of finance, AI, and multilingual data.