Like every successful large enterprise, your organization has likely made remarkable technological progress over the past two decades. The massive IT projects and the correspondingly massive failure rates are largely a thing of the past. But most companies like your aren't poised to thrive in the AI era.
Here is a candid discussion of where you are and where you need to go.
Your technology workforce is optimized for the pre-AI era
Although you could still likely tidy up some inefficiencies and failures in your company's technology landscape, until now, you have likely found success by optimizing along three dimensions:
- Using third-party tools
- Adopting methodologies that create transparency
- Hiring tech generalists at mid-market salaries
Optimization 1: Using third-party tools
Technology has been an enabling factor for traditional sectors such as banking, insurance, health care, and retail. These organizations have profited and maintained market dominance through product improvement, acquisitions, sales and marketing, and operational efficiencies. Unlike tech powerhouses such as Google and Apple, most companies in traditional sectors are better off relying on third-party vendors and tools rather than building custom applications. Your workforce is great at evaluating, buying, configuring, and supporting technology, but they need those third-party companies that have strong engineering competencies.
Optimization 2: Adopting methodologies that create transparency
Because your business is unique, you need to build some custom applications. These projects pose significantly more risk than buying proven solutions. To mitigate those risks, companies like yours have embraced methodologies like agile, SAFe, and scrum. These methodologies reduce risk by creating more predictability and transparency.
Optimization 3: Hiring tech generalists at mid-market salaries
Finally, like most enterprises, you don’t have top-shelf technology talent because you haven’t needed it. The best engineers, product managers, and designers are extremely expensive. They create more economic value at technology companies than in traditional industries. Most companies like yours simply don’t get a good ROI from in-house technology experts.
Example: ETL tools vs. customized DataOps infrastructure
A major cost center of an IT organization is capturing, processing, and storing large amounts of data. Most companies still rely on extract, transform, and load (ETL) tools to populate their data stores. These tools can scale to high volumes of data, and they can be configured and operated by teams of lower-cost technology generalists. Management can supervise the teams’ progress through Jira tickets and reporting. Deploying ETL tools is a process-driven method for building your data processing infrastructure. Historically, this approach to data processing has been more profitable than building a customized DataOps infrastructure.
Tech giants like Facebook, Netflix, Uber, and Google are in the opposite situation. Because their entire business is driven by data, they have invested in top-shelf technology talent to build customized DataOps applications infrastructure. This is an innovation-driven approach for building their data processing infrastructure—and it is the approach you will need to adopt to become an AI-driven organization.
The technology workforce you need for the AI era
Your current workforce is not optimized for AI. Your survival depends on evolving from a process-driven organization to an innovation-driven organization.
When computers start thinking for us
As AI matures, your company will increasingly offload thinking and decision making to computers. Key business drivers will migrate from traditional business processes to technology-driven ones. This revolution has already happened in marketing, where narrative-driven brand messaging has been supplanted by digital marketing. Soon sales effectiveness will be driven by the quality of sales analytics and not the volume of traditional sales metrics such as cold calls. Business processes that rely on a large team of experts today will eventually rely on a small number of experts supported by highly skilled engineers.
Many business processes will become commodities. But processes that create competitive advantages or support unique business offerings will require customized solutions. You will need highly skilled data scientists, data engineers, and product managers to build and support these advances.
Fewer, highly skilled engineers
You can’t ask a group of IT generalists to build the systems that will reinvent your business and expect to stay ahead of the competition. They will simply fall back to their comfort zone of tools, status reports, and Jira tickets. They will try to innovate through processes, and it won’t work.
Modern AI systems (to the degree they exist) are based on state-of-the-art technologies like deep learning and Kubernetes. These systems require fewer engineers to build and support them than traditional, rules-based software applications. These systems are more complex, harder to learn, less mature, and changing rapidly.
Applications based on neural networks are the clearest example of this shift. To create and support a traditional rules-based business application, you need a large engineering team of spec writers, software engineers, and QA specialists, and thousands of lines of code. But if you pack the logic into a neural network with 100 million weights, you can build a complex application by using a small data science and data engineering team.
In the AI era, you will need fewer technology professionals, but they must have more advanced skills. And every person on the team will cost significantly more than you’ve paid for tech talent before.
Analogy with the internet era
A similar workforce shift happened in web application development. In the late 1990s, building a data-driven web application took dozens of server engineers, database administrators, operations engineers, software engineers, and testers.
A decade later, web application frameworks like Ruby on Rails and application hosting platforms like Heroku allowed a smaller group of more-skilled software engineers to create and support a comparably complex solution. The price of skilled software engineers has continued to increase while fewer jobs remain for supporting roles like database and Unix administration.
The solution is obvious–and hard to implement
From a purely economic standpoint, the solution is pretty obvious: you need to realign your recruiting, retention, and compensation structures for the workforce you need, not the workforce you have.
In practice, this transition is hard, a reality that companies began discovering in 2021.
Many companies have attempted to build their data science resources by constraining their compensation package to align with that of their pre-AI workforce. They quickly discover that their top data scientists leave for more lucrative opportunities. Those who remain don’t have the drive or skills to build applications that can solve real business problems.
The top 100 technology companies are at war for AI talent. They are stockpiling the human resources you need. You can’t compete by offering $175,000 to an engineer who can get $275,000 from Microsoft.
Get the free report: Win the War for AI Talent
Fortunately, you have options that don’t involve paying data scientists more than you pay your VP of engineering. Learn the secrets in Prolego’s new report, Win the War for AI Talent.
.png)
.png)
