You can't change what you can't see.
Your legacy system is a black box. We reverse engineer it rule by rule, from your code, and experts verify every finding.
System Intelligence is the understanding every step stands on.
Know what your legacy system actually does.
The whole system, written down and checked: down to the last rule, in plain English, with the exact lines of code behind every finding.
Business logic
- Business rules catalogue, in plain English
- Calculations and decision points
- Exceptions and rare cases, like logic that runs once a year
- Each rule traced to its file and lines
Flows and data
- End-to-end flows, from entry point to outcome
- Data model, entities and schemas
- Interfaces, batch jobs and integrations
- Where each flow reads and writes data
Structure and specs
- Dependency map across repos
- Dead and duplicate code
- Specs and user stories to build from
- Your glossary, confirmed by your experts
Plus a coverage report: 100% of supported code mapped, so you know nothing was skipped.
Plain English on top. The exact code underneath.
Each rule we hand over says what the system does, where it does it and who checked it. Your analysts read the rule. Your engineers jump to the lines.
That's how rare cases surface before a rewrite, like a rule that only runs at year end, or one that exists because a regulator once asked for it.
Claims filed more than 90 days after the incident go to manual review, unless the policy holder is in a declared disaster area.
claims/CLMVAL.cbl · lines 2,431 to 2,458
One map, many reasons to have it.
Teams come to System Intelligence with different goals. The deliverable is the same verified map.
Before a rewrite or migration
Carry every rule into the new system, including the ones nobody remembers.
Then we build itWhen a few people hold the knowledge
Capture what only they know, with them there to confirm it.
Compliance and audit
Show exactly where a regulated rule lives in the code and how it's applied.
Technical due diligence
Read the target's code before you sign: what it does, where the risk sits, what change will cost.
What a diligence coversOnboarding teams and vendors
Give new engineers and partners a map of the system they can trust from day one.
Grounding your coding agents
Serve the same verified map to your agents, so they work from facts.
AI OperabilityAn expert pod, on site or remote.
Subject-matter experts in your domain work alongside our engine. They verify every rule and sit with your people on the terms only they know.
Scoping call
Your system, your goal and what a good outcome looks like.
Grounding sample
One week on a sample of your code, so you see what we can do on your own system.
The full read
Up to a million lines of code a month per pod, depending on complexity.
Handover
The verified map, live in the Swimm platform and ready to build on.
The map stays live, for everyone who works on the system.
When the project ends, the verified map keeps working for you. It lives on in the Swimm platform, in sync with your code, so the system stays easy to read long after the project.
- Analysts look up any rule in plain English.
- Engineers jump from a rule to its exact lines.
- Coding agents query the same map over MCP.
You own what we build; we're there when you want to go further.
Business rules / Claims intake
Claims intake rules
Claims filed more than 90 days after the incident go to manual review.
claims/CLMVAL.cbl · 2,431 to 2,458
Claims from a declared disaster area skip the 90-day review.
claims/CLMVAL.cbl · 2,432
Get our public read of Kill Bill, an open-source billing system, when it's published.
Same method, same deliverables, on code anyone can open. We'll publish what we found and how long each step took. Leave your email and we'll send it when it's live.
github.com/killbill/killbill
About 300,000 lines, mostly Java
Apache 2.0, open to anyone
The read: how proration works when a customer changes plan
The map is where the new application starts.
Application Delivery
We build and deliver the application your business needs on top of the map, with tests that show every rule carried over.
Explore Application DeliveryAI Operability
We make your system ready for agentic development, with the map as verified context for your coding agents.
Explore AI OperabilityFrequently asked questions
Is this documentation?
It's a verified map of how your system works: rules, flows and specs, each traced to its code. You can read it as documentation, search it in the Swimm platform, hand it to your agents and build on it.
How do you reach 100% coverage?
Our engine maps the code deterministically and shows a coverage report before any AI runs. We iterate until every file on a supported stack is mapped, and the report is part of what you get.
Do you need access to our data?
No. We work from code and schemas, never production data, and everything starts under NDA. How your code is handled.
How long does it take?
It depends on the size and complexity of your code. As a guide, one pod reads up to a million lines of code a month. You get dates after the grounding sample.
What do our people need to do?
Give us access to the code, and time with someone who knows the business. They help complete the glossary and answer the questions only your team can.
Know what your system does, before you change it.
30 minutes with our team. Bring the system you have in mind.