Book a discovery call

03 Burnham RNG · United States Shipped

A team building its own financial tools with AI, made production-ready

Burnham RNG's non-developers were shipping internal software fast. We audited it, then put the guardrails underneath: CI, automated review and regression tests on the financial calculators.

Before
Internal project and financial tools built fast with AI coding assistants, with no tests, no CI and no review before a change reached main
After
Every change runs four CI checks and an automated code and security review, and the financial calculators carry regression tests
All green every test passing when the work merged
The work, before and after
Before and afterBefore: Change written with AI, then No tests, then No review, then Straight to main. After: Change written with AI, then Four CI checks, then Automated review, then Calculator tests, then Merged when green. Result: All green, every test passing when the work merged.BEFORE3 OF 4 PAIN POINTSAFTER3 OF 5 AUTOMATEDChange writtenwith AINo testsNo reviewStraight to mainChange writtenwith AIFour CI checksAutomated reviewCalculator testsMerged when greenAll greenevery test passing when the work mergedBefore and afterBefore: Change written with AI, then No tests, then No review, then Straight to main. After: Change written with AI, then Four CI checks, then Automated review, then Calculator tests, then Merged when green. Result: All green, every test passing when the work merged.BEFORE3 OF 4 PAIN POINTSAFTER3 OF 5 AUTOMATEDChange written with AINo testsNo reviewStraight to mainChange written with AIFour CI checksAutomated reviewCalculator testsMerged when greenAll greenevery test passing when the work merged
4
CI checks on every change
Every
pull request reviewed automatically before merge
How the system runs
Input
  • A change, written with AI
ArrAI
CI, review and regression tests
Output
  • Merged when green
  • Caught when a number moves

01 What changed

  • A change that moves a number is caught before it reaches a decision
  • The team keeps shipping at its own speed, with checks underneath it
The full story The problem, how we approached it, and what we built

Burnham RNG builds plants that turn organic waste into renewable natural gas. Its own team, people who are not developers by trade, builds its own internal project tools, using AI coding assistants. We audited the codebase, then made it production-ready.

Built for
Business systems lead
Engagement
Burnham RNG
Sector
Renewable natural gas
Region
United States
Year
2026

02 The problem

The team shipped quickly with AI. What nobody owned was the layer underneath: tests, checks and review.

In a tool that feeds financial decisions, a change that quietly moves a number is the expensive kind of bug.

03 The approach

We started with an audit of the codebase, every finding ranked by severity and handed over with a fix.

Then phase one: CI with four checks on every change, automated review of every pull request for bugs and security, a rules file so the AI assistants follow the codebase's own conventions, regression tests on the financial calculators, a follow-up audit and a workflow that writes new tests. It merged in June with every test green.

04 What we built

  • Code audit, findings ranked by severity, each with a fix
  • CI on every change: typecheck, lint, test, build
  • Automated pull request review for bugs and security
  • Regression tests on the financial calculators
  • A rules file that keeps the AI coding assistants inside the codebase's conventions

Want this kind of result mapped for your business?

Book a discovery call