AI FOMO
Companies today are divided on how much they should adopt AI.
On one hand, you have the startup / solo developer crowd, the advanced AI adopters, who are letting systems like clawdbot run on VMs, allowing full control of the OS via whatsapp commands (I would not recommend this).
Towards the middle of the AI adoption spectrum, you have companies that allow their developers to use enterprise licenses to Claude or ChatGPT, to integrate with their IDE, so that they push features to their code base more efficiently.
For the slow AI adoption group, they are either too budget or talent constrained to integrate these systems, or have already established that AI investments don’t make sense.
Technology leaders recognize where they are on this spectrum of AI adoption. They see the headlines of Ex-Amazon Executives Vibe Coding CRMs. They look at their organization and start thinking “Why am I paying for this product when we could just build it ourselves?”. Competition is healthy. It’s good to see how other companies adopt AI.
However, technology leaders need to be realistic and understand the needs of their business, as their first priority.
If your desire for AI adoption comes from a place of only gaining relevancy, padding your resume, or to signal to your company that you are “making progress for the sake of progress” you will put your company in jeopardy, as you will be blinded by this ambition.
Take for example a small, fictitious healthcare company.
They have pretty low margins and have been historically risk-averse. Sales YOY are relatively flat.
All of their servers are on-prem. Their analytics stack is using SQL Server and Power BI. Things are running smoothly for the most part from a data perspective. The leadership team is generally happy with the dashboards currently in place.
Their CTO reads the headline about the amazon executive vibe coding from earlier. They are inspired by this and want to start using AI. The problem is, the data infrastructure or talent, is nowhere it needs to be to support this. They decide to look into various methods to accomplish this AI initiative.
Since their databases are all on-prem, they look into a data warehousing service like Snowflake, or Fabric (as Power BI integrates well with Fabric). They need a method to replicate data from sql server to this future data warehouse. They find an ELT vendor like FiveTran. They also discover that they have a ton of bespoke python scripts “in production” that are pulling from various healthcare service APIs to create CSVs in SharePoint that their dashboards read from. They want to get away from this practice. To accommodate this, they look into AWS lambdas, which run python scripts to call those APIs and upload data to snowflake on a schedule.
This AI initiative is starting to add complexity to their data infrastructure. They now have to manage the deployments of FiveTran, AWS and Snowflake. They need to hire probably 1-2 people full-time (to do this correctly at least). On top of that, they now incur variable compute charges from each service, depending on the data volume and frequency of updates they need. But all of that work is just to land the data in the data warehouse. They also need to spend significant time converting the existing dashboard sources into Snowflake sources. There are hundreds of dashboards out there! Additionally, to support the AI initiative, they would need to hire a data scientist or machine learning engineer. Using the landed data in Snowflake, they could run models on AWS. That is another cost that needs to be managed.
As you can tell, there is a lot that needs to happen to become “AI ready”. The CTO would need to justify spending hundreds of thousands of dollars and for what benefit exactly? Is the goal to streamline operations with AI so that their providers are more efficient (this could mean layoffs)? Is the goal to show off to potential private equity firms in hopes to sell their business for a higher value? Are there business questions not being answered by the current analytics stack, so they need to reach for advanced classification/regression models? Or are you wanting to adopt AI to pad your resume?
At this point, the healthcare company hasn’t implemented AI. It has implemented an entirely new data platform, a new operating model, and a new cost structure, just to become eligible to attempt AI. None of this is inherently bad. Modernizing data infrastructure is often a necessary evolution. But when the driver is fear of being left behind, rather than a clearly defined business objective, the result is complexity without guaranteed return.
AI is a tool, and like any tool, its value depends on the problem it is applied to. If your current analytics stack already answers the questions your business needs to make decisions, introducing AI will not magically create new value. It will only create new dependencies, new spend, and new operational risk.
Good technology leadership means resisting hype when it doesn’t serve the business.
Technology leaders should absolutely pay attention to how others are adopting AI. Competition matters. Innovation matters. But realism matters more. The right question is not “How do we add AI?” The right question is “What business outcome are we trying to change, and is AI the simplest way to achieve it?”
If you cannot clearly answer that, the initiative is not strategy. It is AI FOMO.