In today’s world, artificial intelligence is everywhere and individuals and businesses alike use it to make their daily tasks easier, quicker, more streamlined and accurate. However, AI is still a mystery in a lot of ways, and because of this, it’s often not being used to its full potential.
For many businesses, the biggest blocker to working with AI is its cost: Why would we want to pay for something expensive when we’re not sure what it is, how it works or what it can do for us?
While there’s no ‘one-size-fits-all’ cost when it comes to AI setup and implementation, we’d like to help give you a clearer idea of what goes into building AI systems so that you can better understand the price tag and make informed decisions.
What is AI?
The first step to understanding how AI can help your business is to understand what it is.
Put simply, artificial intelligence is a technology that can perform tasks in the same manner that humans can. This can range from processing documents to analysing data and predicting outcomes.
Where do we start if we’re interested in implementing AI?
If you’re a business that has only worked with AI in a limited capacity – or indeed, not at all – trying to implement more complex AI systems yourself will be tough. Because it’s an incredibly diverse and malleable technology, it doesn’t come with any form of manual. You also need to make sure that you’re building this technology into your systems in a sustainable way.
> As something you will need to continuously build and maintain, you want to make sure that you’re setting yourself up properly to get the results you want. The most effective way to do this is to partner up with a team of AI experts from the outset.
Software houses sometimes offer AI implementation services – and if you already have a relationship with one, it might seem like tasking them with the new project makes the most sense. However, AI implementation is very different to software development and requires expert knowledge that extends beyond the technical expertise of software development.
As professionals in the field, AI teams have a solid grasp on what the technology is, how it performs in different contexts, and how to build it into existing systems so that it supports and enhances business objectives.
They also understand how new technology will change existing workflows and affect team members, which is a completely different dynamic to consider. Launching AI in your business does not only have technical implications but human ones too: Your team culture will change and you need to know how to handle that for your operations to continue smoothly.
In addition to this, they will also be able to offer you maintenance services, which is important if you want to see on-going results.
What will setting up AI cost me?
As mentioned before, you can never get an exact answer as to how much implementing AI will cost because installation and maintenance fees depend very much on your business’ context, and whether or not you decide to partner with an AI team.
That being said, getting started with AI does not mean you have to break the bank. In fact, for most companies that are looking into AI for the first time, we’ve found that it’s a good idea to choose smaller projects with a clear goal in mind. These can be implemented within a few months and built with scaling in mind. The approach gives you an idea of what AI can look like for your business and help you decide if you want to move all your systems across in the long run.
What other factors will I need to be aware of for budgeting?
AI tooling works with data, so what your current data looks like will influence the work needed to be done. Recalibrating systems looks different in every case, but there are a few factors that could drive the price up or down:
- If you decide to build “once-off” or continuous systems
- How complicated it is to integrate AI with your existing workflows
- The quality and availability of data
- The complexity of the data and the model required to work with it
- Whether or not you need to implement a user interface
Will there be recurring fees once we are up and running?
If you do decide to partner with a team of AI experts, they will offer maintenance services. This is important to continuously see good results from your investment in the tooling. Opting for these services will mean that you will incur recurring expenses. However, these are affordable, and worthwhile to keep up with.
Here are some of the main expenses you can expect on a regular basis:
AI team’s maintenance fee
AI experts will charge you a maintenance fee for their ongoing service. As part of this, they will make sure that the models are performing as expected and improve their accuracy over time.
Subscriptions and licenses
Just like any other software package, AI tooling makes use of existing third-party libraries and tools, and therefore requires a subscription. It is worth it to maintain these subscriptions because external teams are constantly updating and improving the tech – meaning that your systems will always deliver cutting-edge results.
Our team, for example, has developed a library after spending years researching and implementing AI in different contexts. We keep this library up-to-date so that we have access to existing solutions when starting new partnership projects. This means that we can drive implementation costs down because we don’t have to bill clients to build the same features again and again. We charge a subscription fee for the use of our library so that we can continue to build it out.
If your business doesn’t have extensive on-site computing resources to facilitate AI systems, you’ll need to pay for cloud-based infrastructure. Most cloud providers, like Google Cloud Platform (GCP), follow a similar pricing model based on actual usage. Building the kind of elastic infrastructure on-site to handle training and inference, as well as data pipelines, requires a huge capital investment and is not advised in most cases.
Here’s a rough idea of how much these services could cost:
|Recurring service||Minimum cost||Maximum cost|
|AI team’s maintenance fee (if opted for)||R6,000.00||R16,000.00|
|Subscription fees (if used and based on size of company)||R3,000.00||R30,000.00|
|Cloud infrastructure (if used)||R2,500.00||R15,000.00|
*This list is not exhaustive but can help to give you an idea of the recurring expenses that may come up. There are of course other factors that will influence what you pay for depending on your business’ needs.
How can I be sure AI is going to work for my business?
When our team is tasked with implementing AI at a business, it’s important that the client feels comfortable and is 100% on the same page about what is going to happen and how much that is going to cost.
To help both parties feel confident, we’ve developed a three-step process to work through together before any implementation work starts:
- Analysis: We start every project by getting to know the business and understanding its objectives. From this discussion, we can map out potential solutions on our side, that we can then discuss with the client to help them understand what is and isn’t possible. This helps us manage expectations from the get-go. It also provides us with the opportunity to discuss the client’s budget so that we can come to a solution that works for them. Knowing upfront what the bill will look like means they aren’t surprised down the line. This is important for us to build a strong relationship.
- Validation: The next step is focused on validating the proposed AI model. We build something that can be measured against the previously agreed validation metric to prove to the client that the model can deliver the results they want. In other words, we build the model and then test it to see whether or not it will be useful for solving the client’s needs before launching into full-scale production.
- Production: Once everyone is happy, we really get the ball rolling. Once production has started, we launch into taking AI models into production, integrating into the existing systems, setting up machine learning pipelines and creating user interfaces if they are needed. As we go, we make an effort to communicate the work we are doing in as much detail so that the client is actively involved.
Here’s an example to give you a rough idea of how much the process could cost following this three-step process:
|Phase||Minimum cost||Maximum cost|
|Analysis (involving multiple workshops with AI team)||R20,000.00||R50,000.00|
|Validation (processing initial data, building models and presenting progress)||R30,000.00||R100,000.00|
This shows that for as little as R100k, clients can get up and running with AI models in production that can help them take the first step to completely transforming their businesses.
Building AI into your company’s systems is one of the best investments you can make in today’s world. Soon, working with technology that speeds up processes and helps humans deliver accurate results will be a necessity. Getting started now will only put you ahead.
If you’d like to learn more about AI tooling for businesses in general, or the services that Deep Learning Café offers, feel free to get in touch!