Amazon Web Services will spend one billion dollars to send engineers into your office so they can watch you fail at AI for a few weeks and then fix it before you embarrass yourself in front of the board. The program embeds Field Development Engineers with customers who cannot figure out how to make large language models stop inventing facts about their supply chain.
These FDEs will build AI solutions, train your existing team, and leave behind what AWS calls "self-sufficient teams" in a matter of weeks. Self-sufficient means your developers will have just enough knowledge to break something catastrophically three months after the AWS guy leaves and you'll have to call him back at twice the rate.
The billion-dollar investment joins a growing wave of companies realizing that enterprises will pay anything to avoid admitting they hired a data science team in 2023 that still can't deploy a chatbot without it telling customers to eat rocks. AWS figured out it's more profitable to rent you competence than to sell you software and hope you read the documentation.
The embedded engineer model works because CTOs need someone to blame when the AI project fails and it can't be the consultant who was only there for six weeks. It also works because your internal team will nod enthusiastically during the knowledge transfer sessions and then immediately forget everything the moment the FDE walks out the door.
This is the same company that convinced you to lift-and-shift your entire infrastructure to the cloud in 2015 and then charged you forty grand a month to run a Postgres database. Now they'll send a guy named Brett to your office in Plano to explain why your prompt engineering strategy is dogsh*t and your vector database is configured like someone let the intern pick the settings.
Brett will fix it in eleven days and your VP will take credit at the next all-hands.
Photo by on Unsplash
Leave a Comment