About five hands.
That’s how many went up when DHH asked more than 1,000 developers at Rails World on 09.23 who still write meaningful amounts of code by hand.
DHH (David Heinemeier Hansson), the legend in the engineering community who created Ruby on Rails, named 11.24.25, the Opus 4.5 release, as the Kodak Brownie moment of software.
*link to DHH’s presentation
He explained that he has written an average of about 30,000 lines of code per year over the last 21 years. But in August, he wrote 150,000 or five years of code in one month.
That’s 60x his long-run pace.
He’s the first to admit a lot of it is verbose Rust he never reads. Discount it heavily and it’s still an order of magnitude.
“Pencils down”
The Brownie didn’t end painting. It ended painting as the only way to capture an image. Painters moved on, inventing Impressionism and Cubism.
That’s DHH’s frame for 37signals going “pencils down.”
*link to Patrick O'Shaughnessy's tweet with DHH’s slide documenting the imagery boom brought about by the Kodak Brownie, iPhone etc
DHH’s point isn’t that engineers are done. It’s that they evolve from writing code by hand into orchestrating teams of AI agents.
They become makers.
Makers, enabled by Jev
A week before DHH took the stage, Diogo Almeida, a former OpenAI researcher responsible for “co-inventing ChatGPT,” shipped Jev (TypeSafe AI).
A model named after William Stanley Jevons, of Jevons Paradox, the same economist DHH closed on.
*link to Diogo Almeida’s tweet launching Jev
Jev isn’t an LLM.
It’s a semantic decision engine that complements LLMs by making thousands of small calls software needs (is this urgent, who owns it, does a human need to check it) and returning a choice or a score with a confidence level, in milliseconds, for a fraction of a cent.
The LLM keeps the reasoning, staying upstream focused on the complexity.
Jev stays downstream focused on the tactical execution, handling the routing.
Pair makers with the people who know the business
What happens when makers like DHH are paired with Financial Services operators with deep subject matter expertise?
Jev sits on top of your ontology and knowledge graph, so your Enterprise Intuition runs faster and cheaper than on an LLM alone.
*Link to AIWA TOT #34 focused on Enterprise Intuition
Every client request of a wealth management team triaged, scored and routed before a human opens it.
The humans keep the calls that require their discretion, taste and feel.
The intelligence layer handles everything else.
*link to AIWA Intelligence Layer POV
Same question but Sports, Media and Entertainment?
The National Football League (NFL)'s complete universe of footage is indexed by game and date.
Tag every frame by moment and reaction along with a licensed, real-time GIF subscription for fantasy football players could become a $30M+, high-margin business.
You know how much we love to use colorful language and how an image is worth a thousand words…
The tagging is a weekend. The rights work isn’t.
Otherwise, this product would be good to go just in time for this year’s fantasy football playoffs.
You no longer have to wait for these innovations.
They’re here now.
Is your team of makers and operators ready to deploy them?
Humans + Machines. Never Humans vs. Machines.
