Most CEOs are scrambling to use AI in their businesses as a company’s AI capabilities have the potential to become the ultimate competitive advantage. However, CEOs are faced with a difficult question: How to build AI capabilities.
Listening to your internal team can be confusing. Engineers will describe virtues of in-house solutions as it allows them to work on the next cool technology on your payroll. Commercial teams will ask for fast working solutions and won’t mind how much they cost since that cost will likely be booked under tech and not hurt the targets of commercial teams.
Full disclosure, we are running a B2B marketplace so it is in our best interest for companies to work with vendors however there’s enough vendors and transactions in the market so we can afford to be transparent with you and offer a balanced viewpoint.
Understand pros and cons of in-house solutions
The words AI talent war are becoming commonplace. Google query for “AI talent war” returns 1730 results. AI talent is in high demand with tech-giants building AI labs in cities like Redmond, New York, Paris and with VCs willing to invest millions in companies like Curious AI founded by AI researchers that are yet to launch any products. This demand translates into high wages and churn for AI experts. Building inhouse AI talent is expensive.
Beyond wages, AI researchers, like other high in-demand professionals, look for companies with world changing vision and ambition. Your company may be in the business of serving ads but if its mission is to “bring the world closer together” and if it can afford to spend millions on R&D projects that are unlikely to bring any revenue in the short to medium term then it is a lot more interesting than many other companies.
Even if you can get the money and convince prospective employees of your world changing ambition, be prepared to wait. It can take many months to build a well-performing AI solution. It takes time to build and fine-tune AI solutions as a system working based on billions of data points takes time to improve. A faster solution bought from a vendor provides immediate financial and customer satisfaction benefits and potentially enables PR if the solution can be marketed as innovative for its market.
AI, though the term is old, is still an emerging field. In emerging fields like the first days of internet, B2B solutions are mostly missing for niche areas and lack maturity even for more common use cases. So if the company needs a domain-specific solution with high performance requirements, it needs to build one on its own and shoulder the costs.
Understand the advantages of buying an AI solution
If there’s an AI solution in the marketplace offered by a financially stable vendor, that fits your requirements and does not lock you in, then you should probably go for it. Building your AI solution will take time and high costs, making buying a ready-to-use solution attractive.
We laid out the pros and cons of in-house vs procured solution, based on your industry and company size, either in-house solutions or procured solutions should be the primary solution of choice for you.
SMEs or mid-market non-tech companies should stay clear of in-house AI solutions
As we discussed AI solutions require time, money and a strong vision to excite researchers. A non-tech company, especially one that is not a Fortune 500 is unlikely to bring together all three. But this can be a source of strength!
A company with good vendor management skills can tap on the collective resources of best AI vendors in their respective fields. This allows companies access to fast, pay-as-you-go solutions which can help them out-compete their competition. Furthermore, management will focus on the main business not on managing R&D which is notoriously difficult and distracting.
Large non-tech companies or mid-market tech companies may choose to build some AI solutions
Tech companies, even small ones tend to build their own solutions. A high ratio of engineers in any organization pushes the organization to build things, however not all that is built ends up being value creating. The best rule of thumb in any small organization is to Focus Focus Focus. Especially since AI investments are so expensive and time-consuming, a small organization should only make AI investments at the heart of its area of focus when that area is not well-served by vendors.
Surprisingly, it is still a similar situation if you are an AI startup. If the most critical features of your product will be improved significantly with AI techniques, then that can be a worthwhile investment. However, tangential projects need not use AI techniques. Manual efforts, Amazon Turk or just building a simple non-learning system augmented with human labor can solve your problems.
For large non-tech companies, it is again a similar situation. You don’t buy, store and maintain servers today, you use AWS. And you don’t need to clean and process your data, build and maintain your models if that service is provided cost-effectively by a capable AI solution provider.
Largest companies focused on tech need to build their own solutions
Tech giants that have access to significant funding and user data are better off developing their own AI capabilities due to 2 factors:
- It will have positive ROI
- A firm like Google, Facebook, Amazon, Baidu or Walmart has a depth and breadth of data that is almost unmatched outside. The know-how and systems to deal with such large volume of data to produce AI solutions is difficult to find in AI vendors.
- Minor performance improvements in such companies can lead to billions in revenues
- These firms with their huge capital pools can afford to build great AI teams
And this is indeed the case, tech giants over the globe have built AI labs with hundreds of personnel.
Hope you feel better informed about AI investment decisions. Explore AI use cases to understand the concrete areas where AI can add value to your business.