Ten years ago, I co-founded a startup called CubbySpot and it created a platform that helped parents find child care in Toronto.
I remember hiring an offshore team to help with software development, and I spent a lot of time searching, screening, and eventually hiring the software development team. We spent at least $10,000CAD to develop our Most Viable Product (MVP).
Despite good intentions and best efforts, the startup folded and the code was shelved into the vast obscurity of the Cloud Computing void.
This past week, I tried my hand in modern AI-driven software development, or commonly called vibe coding. The first time I really felt the Software Development Life Cycle (SDLC) turn into the AIDLC. And I figured that a good project to do was to revive CubbySpot from the virtual ashes.
I used a combination of Claude Code, Gemini, Antigravity, GitLab and Firebase (I know, my setup is not the most lean). And within a day, I was able to create a prototype of what I think is a new, lovable MVP.
I remembered the time I was negotiating, cajoling, and directing the software development team and contrasting that to what I had just accomplished with a AI Chief of Staff and an AI Coder. The old MVP took at least two months to get off the ground, the new one was under 24 hours.
This is really the future of productivity in software development and it's happening now.
Now, it's not all rainbows and unicorns with vibe coding.
I encountered the "Last Mile" problem with the AIDLC where there are some bugs in the finished product and the time spent debugging the final 20% spent 80% of my time.
Then I realized this "AI Sandwich" in working with this so-called digital twin: AI will really do all of the heavy lifting in coding, but the real work of people will be done before and after that.
There will be a lot of work defining the guardrails of the AI, writing the requirements, verifying alignment with business and customer needs, and prioritizing scope. There will also be a lot of orchestration and determining which AI agent will do what.
Once the code is done, there will be a lot of work directing where the deployment will be, doing the requisite QA, and evaluating if the objectives and goals were met. And of course, the roll out and operationalizing of the said software will be another track to do.
My advice to white collar professionals now is to get ready and prepare for the pre- and post- work that will go along with the work that AI will do. Embrace this AI Sandwich and aim to thrive.
Otherwise, you will be chewed out by the industry.
Top comments (0)