Outsourcing your AI department

SIFR is working with companies in many different industries. Regardless of the industry, many of these companies have a lot in common when it comes to AI-strategies. In particular, one question comes up over and over again: should we build our own AI team or outsource it to an AI development house?

Even though we are an AI development and consulting company, we don’t always recommend that you should be outsourcing as much as possible. Read on to learn when and when not to outsource your AI department.

How to tackle your AI development needs?

Let’s start with why asking these questions is good in the first place? Firstly, the fact that companies have arrived at this question is awesome. It means that many companies have begun to recognize the benefit of using data science to support their decision making. The downside is that they are also realizing that they don’t have a clear understanding of how to get data science capabilities into their company. Do you hire a data scientist or do you build your own AI team? Maybe, it would be best to bring in a high-end consultancy to make sure that whatever machine learning application gets built is built by people who have experience in running successful projects?

Not every organization needs their own data science team. Understanding whether or not you should spend a lot of money on hiring data scientists is crucial for the success of your AI strategy. Even more so, a company needs to understand whether they can hire a data scientists at all? If we look at the data science hiring market then it is currently clearly a workers market. What this means is that good data scientists have their pick of who they want to work for. That usually means that people go for the more exciting employers and by exciting we don’t mean being “startupy” but more companies who have exciting data.

If you are thinking about hiring your own data science team you need to first look at a couple of things to validate that idea:

  1. Do you have the data? Unless you are using it regularly, we can guarantee that your data is in worse shape than you think it is. If you’ve never looked at your operational, customer, or transaction data, then don’t expect quick results. And you shouldn’t hire a data scientist to clean it up.
  2. Can you see more than one machine learning project in your company? If you have just one single question that needs answering, then most likely a highly skilled consultancy is a better option.
  3. Are the problems you are looking to solve ones where people will continually adapt and respond to your actions? If not, then you might not need a team of data scientists to constantly re-train and tune your machine learning models.

Insourcing an AI department

If you have answered yes to all of these questions, then you might think that insourcing or hiring your own team might be the best way forward.

If you decide to go the traditional route of insourcing you need to understand that this is a big decision and most likely won’t give you results quickly. Here is why:

  1. Very often the AI team is positioned in the wrong place inside the company’s structure. Usually they are put near/under business intelligence teams or IT teams. In our experience this is the last place they should be as these teams should be as close to the product development teams as possible. This, however, can create a lot of problems around management’s willingness to have data scientists near their product.
  2. Hiring is hard. Hiring experienced machine learning people is even harder. As mentioned before it is very difficult to get the necessary brains from the market. This WILL take more time than expected.
  3. Your data is most likely a mess. No one likes to admit it, but it’s always the case. Whether the main problem is accessing the data or just the quality (how usable the data is, how many holes there are in the data), or if a person does not have enough experience with working with large datasets all of the time, the progress will be slow.
  4. There is no guarantee that your team will be competent or that the data is enough to provide you with the business value you need from it.

If you manage to overcome these issues and do get a working data science team in-house then it might be one of the most valuable teams in your company. Then it is just a matter of keeping them constantly up-to-date with the latest technology trends and making sure they have enough new data to work with to keep them happy :).

Outsourcing an AI department

Insourcing is not the only option and many companies decide to start off by outsourcing their AI development in the beginning. The reasoning behind this is that taking the “building your own data science team” leap is too risky and costly right now and companies want to try a couple of outsourced projects to validate some ideas. Thankfully there are now AI consultancies who all have machine learning capabilities — or at least claim to. There are many types of machine learning firms out there and as with any vendor, deep due diligence is a necessity.

You’ll want to outsource when you need results quickly but not that often — think of it like a three to four-month project every year. A competent consulting company will set a high bar for the analytics, but will also understand education, adoption, and the difficulty in transitioning to a data-oriented company. The outsourcing company will propose a team, scope out the project, and then deliver to a specific set of outcomes.

For these types of outsourcing projects to be successful, you want experienced consultants with deep domain or industry expertise. You can’t afford to pay for them to learn on the fly. When it works well, you get high-quality results quickly and are happy to pay the fees because the value is so much more.

The downside of outsourcing is the same as with any software development consulting agreement: misaligned incentives, scope creep, inexperienced team members, and high costs. You have to trust that they’ve considered all the details and that they’re not going to sell their new skills to one of your competitors. It’s worthwhile to consider incremental or iterative approaches and to explore alternative contractual arrangements (fixed price, shared revenue, bonuses, etc.).

SIFR hybrid approach

We have developed a hybrid approach to help companies fast-track their AI development. This approach is not for everyone, but it is tailored for companies who have a lot of data and have looked into creating their own teams, but have not yet seen the true value of data science.

Our approach ends up being an insourcing model, but it gets there more quickly by leveraging the expertise of a consulting partner.

The idea is to start out with an outsourced consulting team, but in a longer-term relationship. We don’t simply deliver a single project result. Instead, we start off by setting a strategy, prioritising a set of projects, and beginning to educate the organisation. As the cooperation progresses, our team helps your organisation hire replacements for itself. Over time, the team size remains roughly the same, but fewer and fewer of them are consultants until eventually, it’s an internal team.

The benefit of this hybrid approach is its speed. You get quick results to analytical problems and you get a ready made team at the end of the engagement. The consultants take on much of the difficult change management and infrastructure work while still delivering results.

Whatever approach you might be considering, it is important to decide the route based on data. Look at your products, teams and capabilities and only then decide where to go. At SIFR we believe that any data science initiative is a good one so we wish you good luck in mastering your data.


SIFR is a company focused on empowering your data using artificial intelligence and big data. We pride ourselves by finding real world business value from AI solutions and we help companies set up data science projects, build AI teams and develop an AI strategy. Visit us @ www.sifr.ai