AI-native legacy modernization

Legacy, understood. Then modernized.

Stratum turns a legacy codebase into a validated cloud-native application through one product that deepens in four phases: understand what the system does, plan the migration, protect behavior with generated tests, then transform it, step by step, with evidence on every claim.

Scan a codebase See a sample report

Understand

Plan

Protect

Transform

Understand

Archaeologist

Plan

Migration Planner

Protect

Test Generator

Transform

Modernizer

Archaeologist · Map every module, dependency, table and business process into one knowledge graph with evidence.

Archaeology report

legacy-java-billing

Sample

0score

D · At risk

Technical debt

High

Core revenue components

BillingSubscription engine

Highest risk area

Payment processing

Runtime currency28

Architecture coupling39

Dependency health44

Java 8 · Spring 3 · JDBC/Oracle · SOAP · 48,213 LOC · sample, not a real scan

Built for teams running

Java 8.NET FrameworkOracleSOAPCOBOLTomcatWebLogicSpring 3Struts 1JBossPL/SQLWebSphereEJBPython 2PHP 5MyBatis

How it works

One product that deepens in four phases

Every phase reads from and writes to one knowledge graph per scan, so the plan you approve is the plan that ships.

Phase 1 · Understand

Archaeologist

understand

0score

Runtime

Coupling

Tests

Data

The question: What does this system actually do, and where is the risk?
Deliverable: Archaeology Report: knowledge graph, domains, business processes, modernization score and risks, each with file-and-line citations.
For: CTO, VP Engineering, PE diligence teams

Phase 2 · Plan

Migration Planner

plan

Wave 1 · reportingretain

Wave 2 · taxreplatform

Wave 3 · subscriptionrefactor

Wave 4 · billing + paymentsrearchitect

The question: What do we keep, replatform or rebuild, and in what order?
Deliverable: Target profile, dispositions map, strangler-fig sequence, cost model with scenarios and an exportable playbook.
For: Head of Platform, modernization lead

Phase 3 · Protect

Test Generator

protect

InvoiceJob_ratesProratedSubscriptiongolden

TaxCalculator_appliesRegionalVatunit

PaymentService_capturesIdempotentlyintegration

BillingService_rejectsNegativeAmountunit

The question: How do we change it without breaking what pays the bills?
Deliverable: Characterization test suites (unit, integration, golden) compiled and run in a sandbox, plus a safety score per module.
For: Engineering managers, QA leads

Phase 4 · Transform

Modernizer

transform

public Invoice rate(Subscription s) {

- Runtime.getRuntime().exec("rate.sh " + s.id);

- Connection c = OracleDS.open();

+ RatingResult r = ratingClient.rate(s.id());

+ invoices.save(Invoice.from(r));

}

The question: Can we ship the change safely, step by step?
Deliverable: Verified change sets, cloud artifacts (Dockerfile, Helm, Terraform) and pull requests with migration notes.
For: Platform teams, SI partners

Archaeology report

What you get back from the first scan

A report a CTO can act on and a diligence team can trust: a scored, cited picture of what the system does and where it will fight you.

Sample data · not a real scanRun it on your code

0modernization

D · At risk

Complexity index

Technical debt

High

Core revenue components

BillingSubscription engine

Highest risk areas

Payment processingInvoice generation jobOracle-specific SQL

Lines of code

Files

Modules

Endpoints

Tables

Dependencies

Score components

weight · score

Runtime currency20 · 28
Java 8 (public updates ended), Spring 3.2, Tomcat 7
Architecture coupling20 · 39
3 dependency cycles; billing ↔ payments ↔ subscription
Dependency health15 · 44
log4j 1.2.17, commons-collections 3.2.1, 6 CVEs
Test safety15 · 22
11% of modules have tests; billing has none
Data portability15 · 47
23 Oracle-specific queries, 4 stored procedures
Knowledge risk10 · 58
Bus factor 1 on billing (git blame)
Infra portability5 · 61
Hard-coded hosts in 3 properties files

System summary

86%

A monthly billing platform for subscription products. Subscriptions are rated by a nightly InvoiceJob, taxed through a TaxCalculator with hard-coded regional rules, and settled through a SOAP payment gateway. Revenue-critical logic is concentrated in two modules that have no automated tests.

src/main/java/com/acme/billing/InvoiceJob.java:40–188src/main/java/com/acme/billing/TaxCalculator.java:12–96src/main/java/com/acme/payments/PaymentGatewayClient.java:55–131

Platform

Built to be trusted with the system that pays the bills

A deterministic analysis engine does the structural work. Models only summarize, label and propose on top of it, and everything they say is cited.

Evidence on every claim

Every summary, domain, risk and process points at file paths and line ranges and carries a confidence. Unparsed files are listed, not hidden.

Invoices are rated by a nightly job that applies regional VAT rules before persisting.

billing/InvoiceJob.java:40–188billing/TaxCalculator.java:12–96

confidence0.86

Runs anywhere

Use the hosted service or run the VPC runner inside your own cloud so source never leaves your network. Same product, same report.

Stratum Cloud

isolated per-org storage, encrypted at rest

VPC runner

engine, worker and sandbox inside your cloud

Your code is never used to train models. Secrets are encrypted with Fernet and never returned unmasked.

Model-agnostic

Bring your own keys or run local models. Route each workload to the provider that fits, from frontier models to Ollama and vLLM on your hardware.

0provider kinds, routed per role

AnthropicOpenAIAzure OpenAIGoogle GeminiGoogle Vertex AIAWS BedrockMistralCohereGroqTogetherFireworksOpenRouterxAIDeepSeekPerplexityHugging FaceCerebrasSambaNovaAlibaba QwenMoonshotZhipuOllamaLM StudiovLLMOpenAI-compatible+ offline mock

Bulk, synthesis, planning, coding, chat and embeddings can each use a different provider, with fallbacks and a daily budget.

Pricing

Start with one scan. Grow into a program.

Pay per scan while you evaluate, or move to a yearly plan when the whole team needs the full pipeline. No monthly toggle, no seat games.

Free

open source or under 50K lines

Try the Archaeologist on a small or open-source codebase.

Full archaeology report
Knowledge graph explorer
Ask-the-codebase chat
Bring your own model keys
Community support

Scan a codebase

Team

Business

$25K

per year

Unlimited scans and the full four-phase pipeline for one team.

Everything in Team
Unlimited scans and re-scans
Test Generator with sandbox runs
Modernizer change sets and PRs
Role-based LLM routing and budgets

Start with Business

Enterprise

$100K+

per year

For portfolios, regulated environments and SI partners.

Everything in Business
VPC runner in your own cloud
SSO and SCIM
Audit log and data retention controls
Dedicated support and onboarding

Contact sales

Assessments

Fixed-scope engagements, delivered on the platform

Priced per system. Every deliverable is produced by the product, so you keep the workspace after the engagement ends.

PE & M&A

Diligence Brief

$10–25K

A read-only archaeology of a target's codebase in days: score, risks, key-person exposure and a modernization estimate for the deal team.

Engineering leadership

Modernization Assessment

$40–75K

Full archaeology plus a target profile, dispositions and a costed strangler-fig sequence with scenarios, reviewed with your architects.

Modernization programs

Program Playbook

$75–100K+

The assessment, a generated safety net of characterization tests and the first verified change sets, packaged as a program plan.

Scope an assessmentAll prices in USD. Annual plans include onboarding.

FAQ

Questions buyers ask before the first scan

Security, accuracy, scope and what the numbers mean.

Legacy, understood. Then modernized.

Start with an Archaeology Report. Decide with evidence.

Upload a ZIP or connect a repository. The first report on a small or open-source codebase is free.

Scan a codebase See pricing

AI-native legacy modernization

Legacy, understood. Then modernized.

Scan a codebase See a sample report

Understand

Plan

Protect

Transform

Understand

Archaeologist

Plan

Migration Planner

Protect

Test Generator

Transform

Modernizer

Archaeologist · Map every module, dependency, table and business process into one knowledge graph with evidence.

Archaeology report

legacy-java-billing

Sample

0score

D · At risk

Technical debt

High

Core revenue components

BillingSubscription engine

Highest risk area

Payment processing

Runtime currency28

Architecture coupling39

Dependency health44

Java 8 · Spring 3 · JDBC/Oracle · SOAP · 48,213 LOC · sample, not a real scan

Built for teams running

Java 8.NET FrameworkOracleSOAPCOBOLTomcatWebLogicSpring 3Struts 1JBossPL/SQLWebSphereEJBPython 2PHP 5MyBatis

How it works

One product that deepens in four phases

Every phase reads from and writes to one knowledge graph per scan, so the plan you approve is the plan that ships.

Phase 1 · Understand

Archaeologist

understand

0score

Runtime

Coupling

Tests

Data

The question: What does this system actually do, and where is the risk?
Deliverable: Archaeology Report: knowledge graph, domains, business processes, modernization score and risks, each with file-and-line citations.
For: CTO, VP Engineering, PE diligence teams

Phase 2 · Plan

Migration Planner

plan

Wave 1 · reportingretain

Wave 2 · taxreplatform

Wave 3 · subscriptionrefactor

Wave 4 · billing + paymentsrearchitect

The question: What do we keep, replatform or rebuild, and in what order?
Deliverable: Target profile, dispositions map, strangler-fig sequence, cost model with scenarios and an exportable playbook.
For: Head of Platform, modernization lead

Phase 3 · Protect

Test Generator

protect

InvoiceJob_ratesProratedSubscriptiongolden

TaxCalculator_appliesRegionalVatunit

PaymentService_capturesIdempotentlyintegration

BillingService_rejectsNegativeAmountunit

The question: How do we change it without breaking what pays the bills?
Deliverable: Characterization test suites (unit, integration, golden) compiled and run in a sandbox, plus a safety score per module.
For: Engineering managers, QA leads

Phase 4 · Transform

Modernizer

transform

public Invoice rate(Subscription s) {

- Runtime.getRuntime().exec("rate.sh " + s.id);

- Connection c = OracleDS.open();

+ RatingResult r = ratingClient.rate(s.id());

+ invoices.save(Invoice.from(r));

}

The question: Can we ship the change safely, step by step?
Deliverable: Verified change sets, cloud artifacts (Dockerfile, Helm, Terraform) and pull requests with migration notes.
For: Platform teams, SI partners

Archaeology report

What you get back from the first scan

A report a CTO can act on and a diligence team can trust: a scored, cited picture of what the system does and where it will fight you.

Sample data · not a real scanRun it on your code

0modernization

D · At risk

Complexity index

Technical debt

High

Core revenue components

BillingSubscription engine

Highest risk areas

Payment processingInvoice generation jobOracle-specific SQL

Lines of code

Files

Modules

Endpoints

Tables

Dependencies

Score components

weight · score

Runtime currency20 · 28
Java 8 (public updates ended), Spring 3.2, Tomcat 7
Architecture coupling20 · 39
3 dependency cycles; billing ↔ payments ↔ subscription
Dependency health15 · 44
log4j 1.2.17, commons-collections 3.2.1, 6 CVEs
Test safety15 · 22
11% of modules have tests; billing has none
Data portability15 · 47
23 Oracle-specific queries, 4 stored procedures
Knowledge risk10 · 58
Bus factor 1 on billing (git blame)
Infra portability5 · 61
Hard-coded hosts in 3 properties files

System summary

86%

src/main/java/com/acme/billing/InvoiceJob.java:40–188src/main/java/com/acme/billing/TaxCalculator.java:12–96src/main/java/com/acme/payments/PaymentGatewayClient.java:55–131

Platform

Built to be trusted with the system that pays the bills

A deterministic analysis engine does the structural work. Models only summarize, label and propose on top of it, and everything they say is cited.

Evidence on every claim

Every summary, domain, risk and process points at file paths and line ranges and carries a confidence. Unparsed files are listed, not hidden.

Invoices are rated by a nightly job that applies regional VAT rules before persisting.

billing/InvoiceJob.java:40–188billing/TaxCalculator.java:12–96

confidence0.86

Runs anywhere

Use the hosted service or run the VPC runner inside your own cloud so source never leaves your network. Same product, same report.

Stratum Cloud

isolated per-org storage, encrypted at rest

VPC runner

engine, worker and sandbox inside your cloud

Your code is never used to train models. Secrets are encrypted with Fernet and never returned unmasked.

Model-agnostic

Bring your own keys or run local models. Route each workload to the provider that fits, from frontier models to Ollama and vLLM on your hardware.

0provider kinds, routed per role

Bulk, synthesis, planning, coding, chat and embeddings can each use a different provider, with fallbacks and a daily budget.

Pricing

Start with one scan. Grow into a program.

Pay per scan while you evaluate, or move to a yearly plan when the whole team needs the full pipeline. No monthly toggle, no seat games.

Free

open source or under 50K lines

Try the Archaeologist on a small or open-source codebase.

Full archaeology report
Knowledge graph explorer
Ask-the-codebase chat
Bring your own model keys
Community support

Scan a codebase

Team

Business

$25K

per year

Unlimited scans and the full four-phase pipeline for one team.

Everything in Team
Unlimited scans and re-scans
Test Generator with sandbox runs
Modernizer change sets and PRs
Role-based LLM routing and budgets

Start with Business

Enterprise

$100K+

per year

For portfolios, regulated environments and SI partners.

Everything in Business
VPC runner in your own cloud
SSO and SCIM
Audit log and data retention controls
Dedicated support and onboarding

Contact sales

Assessments

Fixed-scope engagements, delivered on the platform

Priced per system. Every deliverable is produced by the product, so you keep the workspace after the engagement ends.

PE & M&A

Diligence Brief

$10–25K

A read-only archaeology of a target's codebase in days: score, risks, key-person exposure and a modernization estimate for the deal team.

Engineering leadership

Modernization Assessment

$40–75K

Full archaeology plus a target profile, dispositions and a costed strangler-fig sequence with scenarios, reviewed with your architects.

Modernization programs

Program Playbook

$75–100K+

The assessment, a generated safety net of characterization tests and the first verified change sets, packaged as a program plan.

Scope an assessmentAll prices in USD. Annual plans include onboarding.

FAQ

Questions buyers ask before the first scan

Security, accuracy, scope and what the numbers mean.

Legacy, understood. Then modernized.

Start with an Archaeology Report. Decide with evidence.

Upload a ZIP or connect a repository. The first report on a small or open-source codebase is free.

Scan a codebase See pricing

Legacy, understood. Then modernized.

One product that deepens in four phases

Archaeologist

Migration Planner

Test Generator

Modernizer

What you get back from the first scan

Score components

System summary

Built to be trusted with the system that pays the bills

Evidence on every claim

Runs anywhere

Model-agnostic

Start with one scan. Grow into a program.

Free

Team

Business

Enterprise

Fixed-scope engagements, delivered on the platform

Diligence Brief

Modernization Assessment

Program Playbook

Questions buyers ask before the first scan

Is my source code secure? Where does it go?

Can it run inside our VPC?

Which languages and stacks does it understand?

How accurate is it?

What does the modernization score mean?

How are the tests generated?

Can we use it for private-equity diligence?

Which models can it use, and who pays for them?

Start with an Archaeology Report. Decide with evidence.

Legacy, understood. Then modernized.

One product that deepens in four phases

Archaeologist

Migration Planner

Test Generator

Modernizer

What you get back from the first scan

Score components

System summary

Built to be trusted with the system that pays the bills

Evidence on every claim

Runs anywhere

Model-agnostic

Start with one scan. Grow into a program.

Free

Team

Business

Enterprise

Fixed-scope engagements, delivered on the platform

Diligence Brief

Modernization Assessment

Program Playbook

Questions buyers ask before the first scan

Is my source code secure? Where does it go?

Can it run inside our VPC?

Which languages and stacks does it understand?

How accurate is it?

What does the modernization score mean?

How are the tests generated?

Can we use it for private-equity diligence?

Which models can it use, and who pays for them?

Start with an Archaeology Report. Decide with evidence.