When you have a problem that you want to solve with AI, you are faced with two main choices: use an existing solution provided by someone else or build your own. In reality, there are also several shades of gray in between these two extremes as well. Below we discuss a few approaches along with their pros and cons.
If you’re trying to decide which of these approaches is best for you, please don’t hestiate to get in touch with us at Inspired Cognition, and we can provide some insight on which is best for your use case.
Off-the-shelf API-based Solution
An off-the-shelf API-based solution is one hosted on a cloud service that you can access programmatically. These solutions can further be divided into two categories: application-specific and general-purpose.
Application-specific API Providers
Application-specific API providers provide a gateway to specific AI-based applications, such as image classification, text classification, speech recognition, or machine translation. These APIs have the advantage of being very easy to use, but the disadvantage of having limited flexibility and being less customizable. You may also need to take care regarding terms of use regarding private data, although many providers also allow on-premise hosting on your servers.
Notably, these APIs also have a varying degree of customization available. Many of them are meant to be simply used out-of-the-box for specific tasks, some of them allow a degree of customization using your own data, and some are designed to low-code or no-code AI development solutions that provide a relatively large degree of flexibility.
If you’re interested in this approach, you can find a list of some of the most popular API-based solutions for various popular tasks on our API-based AI Application List.
General-purpose API Providers
In the past few years, a number of companies have started offering general-purpose AI platforms that allow you to build and deploy solutions for new AI tasks by describing the tasks in a high-level way. These platforms are often very flexible within certain constraints (e.g. they can handle most tasks expressed in English), but also can be somewhat difficult to control.
For more detail, see the separate page on General-purpose AI Platforms, where we list providers and give an idea of how to use them.
In-house Development
If you are a developer or have a team of developers at your disposal, you can develop the system yourself. This is the most flexible approach and can end up being cheaper in the long run as you scale usage larger, but it is also the most costly in terms of initial investment of time and money. In addition, you will need to consider how to deploy your model in a way that is scalable and robust.
If you are interested in pursuing this approach, you can either take a look at our AI Model Prototyping guide that gives you an idea of where to get started, or rely on a trained and experienced developer. If you would like to get in touch with us at Inspired Cognition about custom development projects led by world-class AI experts, please contact us using the contact button above!
Next Steps
Once you have an AI solution, the next step is to deploy it as part of your larger product or service.