Day one
A codebase we'd never touched. 30+ microservices, business logic spread across services no one had fully mapped.
Hard deadline
New market live in 30 days. The traditional path — weeks of briefings, manual tracing, a team of 10–15 engineers — wasn't an option.
Here's what we did instead.
30
Days to market
vs 90–120 standard
32
Services
mapped autonomously
40%
Lower cost
vs traditional dev
0
Architecture briefings
agent mapped the architecture
A system we didn't build. A market we had to launch in 1 month.
MyCredit runs on a distributed .NET 8.0 ecosystem with 30+ independently deployed microservices. We had not originally built this system. Critical business logic was spread across separate services — registration, KYC verification, fraud marking, audit logging. A change in one flow could silently cascade into failures elsewhere.
The real bottleneck in systems like this isn't writing the code. It's understanding what touches what — before touching anything.
Understanding the system was the hardest part.
What made this complex:
- 30+ microservices, distributed logic
- Cross-service cascade risk
- Unfamiliar .NET 8.0 codebase
- KYC, compliance, fraud rules spread across services
- High cost of release mistakes at this scale
One change to a registration step could affect customer data flows, fraud checks, audit events, or validation logic elsewhere in the system.
Traditional approach: weeks of architecture walkthroughs, heavy reliance on client-side leads, manual service tracing, slow planning before implementation even starts.
That is where large-system delivery becomes both slow and expensive.
Spec-driven development: the agent reads the system.
Instead of unpacking the system service by service, we used AI agents to work across the entire architecture simultaneously.
Semantic Mapping
AI agents indexed all 32 services and built a cross-service dependency map.
Logic Orchestration
Changes defined at business-flow level — DRP documents, KYC uniqueness, payment adapters — executed across all services in parallel.
Consistency Validation
Agents validated every change against fraud marking, audit events, and customer state transitions before deployment.
Agent log
- 32 services
- 47 contracts generated
- 112 tests written
- Minimal client involvement
From zero architectural familiarity to stable release. In 1 month.
The client did not have to freeze lead engineers for weeks to explain the system. Fewer meetings. Less manual tracing. Less coordination drag. Faster path to implementation.
METRIC
Time to market
Team size
Architecture briefings
Cost
Bugs at launch
Next market
TRADITIONAL
90–120 days
10–15 engineers
Weeks
High discovery overhead
High risk
Full cycle again
ARTJOKER
AGENTIC SDD
30 days
2–3 engineers
0
~40% lower
Stable release
Near-zero marginal effort
When architecture complexity goes beyond what a team can efficiently reason through manually — Spec-Driven Development (SDD) is a faster, leaner, and safer path to delivery.
Right fit:
We also help with:
- Architecture understanding in unfamiliar codebases
- Cross-service delivery orchestration
- Release-safe implementation at scale
- Reducing delivery overhead on complex systems
Kashcheiev Maksym
Head of Business Development
contact us:
USA
Estonia
Ukraine
Norway
