Maintainability
A codebase teams can actually understand
Normal changes no longer require navigating a maze of disconnected services. The codebase is logical, understandable, and easier to grow with the business.
Cadence Case Study
Cadence’s prior-authorization platform is how they deliver their service. Normal changes had become slow, costly, and risky. We learned how the business worked. We decided more patches would make things worse. We rebuilt the platform in about six weeks on a foundation that is easier to maintain. We still support it.
About 5 minutes. No sales call required.
Built with Cadence, a medical prior-authorization service for hospitals, clinics, and payers.
“Cadence” is a pseudonym used at the client’s request. The engagement, work, and outcomes described here are real.
~6 Weeks
Full platform rebuild
90%+
Lower monthly hosting costs
Days → Minutes
Many support needs: days of waiting → about 30–45 minutes
Ongoing
Senior support continues today
Support timing describes typical experience after the rebuild and is not a contractual service-level agreement.
Client
Cadence — a medical prior-authorization service for hospitals, clinics, and payers.
Industry
Healthcare technology / medical prior authorization.
Challenge
A working custom platform had become slow, costly, and risky to maintain or expand.
Solution
A complete rebuild on a right-sized, easier-to-maintain setup, followed by ongoing senior technical support.
Outcome
Faster development, improved reliability and usability, much lower hosting costs, and continued involvement from the senior team.
Cadence handles medical prior authorizations for hospitals, clinics, and payers. Their service uses AI-supported workflows plus expert staff for phone calls, faxes, paperwork, research, and follow-up. The platform puts all of that work in one coordinated process.
This is not a side tool. The app is how the service is delivered. Requests usually must turn around in 24 to 48 hours. When software for a critical workflow becomes hard to change, the whole business feels it.
Cadence was right to build custom software. Generic tools could not give them the workflow, control, ownership, or flexibility they needed. Custom healthcare software was the right choice — and it stayed the right choice through the rebuild.
Custom development was not the mistake. The first version turned a founder’s idea into a real business. The later question was not whether to build custom software — it was how to structure it.
The first version was built by an overseas team found through a trusted referral. It used React, AWS, and many small connected services (microservices). It worked at first and helped launch the business.
Problems showed up later — not at launch, but as Cadence tried to maintain and grow the platform. Every fix, feature, and release cost more than it should. That is technical debt: a complexity tax on every future change.
Small changes took a disproportionate amount of engineering effort.
Bugs were difficult and slow to trace across disconnected services.
Hosting costs kept rising.
More developers and junior handoffs made the codebase harder to understand.
Communication required extensive written specifications.
Time-zone and healthcare-domain differences added friction.
“Anyone can get software to work once. The question is what it costs to change it on the hundredth or thousandth day.”
Eventually Cadence stopped requesting changes and built manual workarounds inside their own platform — because changing the system had become too hard.
On paper, the microservices design looked advanced and scalable. In practice, it solved a problem Cadence did not have. It was built for huge scale, not for Cadence’s real business and small engineering team.
Architecture should let a small team ship reliably and affordably — not copy Netflix. Fancy is not the goal. Fit is the goal.
Cadence first hired 21 Solutions to understand, host, support, and slowly improve the existing app — not to sell a rebuild. Our first instinct was to make the current system work.
During review, that plan hit reality. Even a simple field change meant tracing logic across many apps, controllers, services, and databases. A minor change could use up most or all of a small monthly support budget because of the broken-up architecture.
We told Cadence honestly that taking money to keep patching the platform would not be a responsible use of their budget. Senior tech judgment includes telling a client what they do not need.
“Architecture decisions are business decisions in technical clothing.”
The tension was real. Cadence had already spent years building and running the platform. Starting over felt risky. Sunk cost is real, and no one wants to walk away from years of investment.
But the cost to build is not the same as the cost to own over time. After looking at the real cost and ability to keep maintaining the platform, Cadence concluded that rebuilding was cheaper and more responsible — the option they could actually carry out.
Before choosing a tech stack, 21 Solutions worked to understand how the business actually ran. A development partner should understand the business before picking an architecture.
The right setup is not the one that sounds most advanced. It is the one that lets the business operate, change, and grow reliably.
With that understanding, the team chose a straightforward setup — a well-structured monolith (one unified app) that matched Cadence’s stage, team, workflow, and development needs. This was not saying monoliths beat microservices everywhere; it was the right fit here.
A platform that had been running and changing for about two years was rebuilt and relaunched on a clean, well-structured monolith in about six weeks, using the team’s established app patterns. The work went well beyond rewriting screens.
Maintainability
Normal changes no longer require navigating a maze of disconnected services. The codebase is logical, understandable, and easier to grow with the business.
Reliability
Cadence gained more confidence that requests moving through the platform would finish correctly. Previously troublesome fax workflows became dependable after the transition.
User experience
Information and AI-supported responses became easier to sort, understand, and navigate for the specialists doing the work.
Support speed
Requests that previously waited days can often be handled through direct communication in about 30–45 minutes, depending on complexity. This describes typical experience, not a contractual service level.
Hosting cost
The right-sized setup and hosting migration reduced monthly hosting expenses by more than 90%.
Senior technical support
Cadence no longer needs to translate every business need into a detailed technical product document. They can share the business need and rely on 21 Solutions to determine the right implementation.
Founder advice
“Find someone who will tell you what you don’t need.”
Technical complexity is easy to sell because complexity is billable. A good tech partner asks about the business before picking an architecture — and recommends less when less is what the business needs.
“21 Solutions is our in-house engineering team at this point. I can communicate the business need, and they understand what has to happen technically to execute it.”
A platform does not have to be fully broken to become a business risk. If normal improvements are slow, costly, or unpredictable, 21 Solutions can help you decide whether to stabilize, modernize, or rebuild responsibly.