You’ve come up with a brilliant idea for your startup. You want to start taking the idea to the next stage but are unsure if you are ready or close enough to start building something profitable and purposeful for your business.
Finding the right resources about what you need to have in place to begin your business is difficult because much of the information you are reading or listening to applies to tech founders or those who have considerable experience developing tech products.
AI is different in that it has many more resources necessary to start. This article will help you review some of the key things you must have in place for a successful start.
You have a large amount of domain or industry knowledge.
Making mistakes executing an AI project is a slow fix, unlike any other tech development. That’s because data mainly drives AI, and you’ll need the right data to be profitable. You also need the right data to be successful with the problem you aim to solve. Collecting or owning the wrong data can set you back.
You can best avoid or best minimize this error by focusing on areas where you have the greatest expertise and skills. You’ll have an advantage that will allow you to quickly understand the industry’s pain points that keep repeatedly coming up. You can identify ideas to solve the problem.
The learning curve for some sectors and industries can be long, and when you pick a business idea that falls out of your expertise, you need more capital and resources to invest. You’ll be spending time and money to learn before you can start to make headway.
If you are considering entering regulated industries like healthcare or transportation, you should know or at least have someone on your team or advisor who knows the industry inside and out. Regulated industries typically bring a lot of red tape and barriers to keep things a status quo. If you don’t know well ahead, you could be spending time going in circles without making any progress.
You have a single, small use case that will add value or solve a problem from the start
You may have a big vision of what the final product should be. Even though your technology will constantly evolve over time, we often want to build that final product from the start. However, it’s not the most practical way to design a product. More so for an AI product.
You want to start small because you’re new to using AI for the first time. Starting small will give you the opportunity to learn. Consider this your capital investment in learning to give you the answer on how you can build the rest of the product. Your basic product will allow you to capture how well AI can solve problems in the real world with real interaction with humans. Not with existing data. You will also learn if you have sufficient data to address the problem and what you need to do to hit your next milestone.
Your initial feature should include the minimum required to make a difference in people’s lives and enough to solve a problem you are addressing. The people who will gravitate towards your product’s first version are those who either tried solving the problem already with no success. They desperately want a solution that could help them right away without all the fancy nice-to-have but not must-have features that can come later. More importantly, if it’s a commercial application, you want to be validating there is a commercial opportunity as soon as you can.
You have a large amount of data to support your initial use case
I know I keep repeatedly talking about data. But for that initial use case or feature that you want to build, you need to have enough data for people to start seeing success. The amount of data is relative to the problem you are solving. It is also based on what machine learning techniques you are applying. Some techniques can help reduce the amount of data you need.
It’s a trial and error to find what AI model gives the best results. When it has the potential to fail repeatedly, you’ll need a solution in place. This often means having human intervention to ensure that the system operates as it should under various circumstances or doing manual activities to fill in the gap.
Remember, you are building a data powerhouse company first. Recognize that you may not have all the data you need from day one. Start preparing for the collection of more data as you grow and better understand your problem space.
You have sufficient capital to start and keep going
Many tech companies can be built all in a day. You hear about weekend projects that are suddenly up and running in hours with thousands of users. However, building an AI product is considered deep tech because AI is in its infancy. Deep tech requires intense capital and time to begin reaching commercial success.
You’ll see at the time of writing AI has been mainly for those who are research and development (R & D)companies, like universities or non-profits R & D start-ups. They strategically designed their business to focus on building and raising capital without the pressure of meeting commercial success soon. The bulk of their budget goes towards developing experiments without investors breathing down their necks, wanting for early returns.
Your initial use case and getting the initial sales can be stepping stones to bringing new cash into your business.
You can find the talent at various stages to support you with your development
Talent is really important for AI-led product development. This holds true, especially if you are a non-technical person or someone without a background in programming. The talent is very limited, given by the popular demand for AI talent and the competitive behaviours that companies do to attract these talents. I’ve heard companies turning workplaces into shopping malls like an all-you-can-eat buffet of services, from free gourmet meals to hairdressers and yoga studios. All in the effort to win talent that is hard to come by.
For some talent, you are attracting them with the best competitive salary and for others, they want to work with you given that you have a solid mission and ambitious goal that can excite someone to help fulfill that mission.
In any big endeavour, you want to sell on the mission and vision to attract talent and bring people eager to support you in your product development.
You have a basic understanding of AI to lead the project
Don’t expect to be able to master and learn everything you need to know about AI and become a machine learning expert. The learning process will take time as you develop the fundamentals to help you better understand and learn the field enough to take on programming projects.
While there are tools to help democratize AI and allow you to learn the fundamentals and the steps to build an AI model, it will take years of practice and consistent learning. The average time to begin to start calling yourself an expert can be 3 years, but after that, it could be more than 3-5 years of regular learning.
You don’t have that much time.
So, you would want to bring someone with the training and capabilities to develop your first product. You want to at least gather the basics to help articulate your needs and set a direction for where you want to go and how you will solve the problem.
When you’re ready
With the five things on top of your mind, you’re ready to build your version one with this framework. With domain expertise, data and capital to solve the simplest problem and a solid team, you will start to see that you are well prepared to develop a product that will make an impact.