Back to case studies
Agentic Modernization of a Legacy Accounting Product
Client name withheld · Finance
Agents
48
Tool calls analyzed
~868,000
Tool error rate
2.1%
A legacy C++ accounting product moving to .NET MAUI and PostgreSQL, one story at a time: a 48-agent engine extracts the specification, then plans, builds, tests and reviews each story, and learns from its own mistakes.
What changed
Before
Product
Legacy C++
accounting product
After
Specify
Spec extraction
9 stages
Plan
Architect
plan gate on Opus
Build
Developer and QA agents
.NET MAUI · PostgreSQL
Review
Review council
5 seats
Learning loop
The legacy product's behavior becomes a specification first. Then every story goes through a plan gate, a build, tests and a five-seat review, and every mistake becomes a rule.
What we built
The engine moves a legacy C++ accounting product to .NET MAUI and PostgreSQL. A tech-lead orchestrator ships one story at a time and loops until the dependency graph is empty.
Key Features
Spec extraction in 9 stages, among them behavior, data and UI lenses, a reconciler, an architecture mapper, a spec auditor and a backlog publisher.
A dev loop where an architect plans, a plan validator gates the plan, developers build and QA agents test.
A five-seat review council that checks architecture, security, legacy parity, UI accuracy and memory leaks.
A learning loop that turns mistakes, gaps and defects into rule changes, with a lesson journal to catch repeats.
A cost-aware model mix: Haiku for mechanical checks, Sonnet for most work, Opus only for the plan gate.
Results
48 agents, 19 commands, 9 workflows and 21 CI jobs in one engine.
About 868,000 tool calls analyzed, at a tool error rate of 2.1%.
The legacy product's behavior, data and UI captured as a specification and published as a backlog.
Stack
Next step
Let’s look at your SDLC together.
45 minutes with a senior engineer: your setup, the risks, what to fix first. No sales team.