Case Study
A Regen Story
Built before standards existed. But it became the backbone of our relationship with Itaú.
And with them went the knowledge of what actually mattered.
100 endpoints documented. We only needed 10.
The 10 we needed were the most complex.
Groovy to Go. Different paradigms.
Groovy runtime. Cannot be patched.
def calculateTax(amount) {
// TODO: review this later
return amount * 0.0825
}8.25% — Why that specific value? Nobody could tell us.
Signal lost in noise.
Uncertainty doesn't shrink with headcount.
The running system IS the specification.
$ curl legacy-api/calculate?amount=1000
{
"tax": 82.50,
"total": 1082.50
}
# → Golden test case capturedEach response becomes a golden test case.
def calculateTax(amount) {
return amount * 0.0825
}func CalculateTax(amount float64) float64 {
return amount * 0.0825
}The AI figured out that magic tax rate just from input-output pairs.
Like-for-like means bugs too. We weren't fixing the system.
We were replicating it exactly.
Testing: /calculate?amount=1000
Legacy: { tax: 82.50, total: 1082.50 }
Modern: { tax: 82.50, total: 1082.50 }
✓ EXACT MATCH
Testing: /invoice/lookup/INV-2024-001
Legacy: { status: "paid", amount: 5000 }
Modern: { status: "paid", amount: 5000 }
✓ EXACT MATCHThe running system was the specification all along.