IntersogRiding Technology's Wave®

Services / Custom Software and Modernization

Tests first. Then agents. Then the new system, in your repository.

Before any agent changes a line, Intersog captures what your system does today as tests. Every phase ends when the same tests pass on the new code. You keep the tests whether or not we continue.

The route

From the audit to a system on supported versions

  1. 01

    10-day Readiness Audit

    Map, coverage gaps, risk list, phase plan with an AI-spend estimate.

  2. 02

    Test harness

    The tests that record what the system does today, running in your CI.

  3. 03

    Phases

    Each phase at a fixed price. Small pull requests, each merged by a person.

  4. 04

    Regression window

    Kept in scope after cutover. Regressions are traced and fixed.

  5. ✓

    Done

    The same tests pass on the new system. You keep the tests.

First step

The Codebase Readiness Audit

Ten business days. A written plan you keep, with or without us. We skip it only where our team already works in your code.

Day 1

Read access, scoped to the repository.

Days 2 to 9

One senior engineer and the agent toolkit read the code and talk to your team.

Day 10

The written plan and a readout.

You get

A map of the system and its dependencies. The test-coverage gaps. A risk list, in order. A phase plan with the AI-spend estimate for each phase.

Priced at a fixed fee, quoted on request. If you continue with us, the fee is credited against the first phase.

After the audit

Three ways forward

Legacy Modernization

Upgrades the system where it stands. End-of-life frameworks, runtimes, and dependencies move to supported versions. Then come tests in CI, a code map, and conventions, so AI agents can work on it safely.

Done when the system runs on supported versions and the test suite is green in your CI.

Agentic Modernization

Moves the system: a port, a re-platform, or a rewrite. One senior engineer directs AI agents inside the test harness (the tests that record what the system does today) and merges the work in small pull requests.

Done when the same tests pass on the new system and the agreed target is met.

Monthly engineer

One senior engineer works on your roadmap with you, month by month. For a system that needs steady, owned change rather than one move.

Done when the monthly outcomes agreed in advance are met.

New software is built the same way: scope bounded first, then a fixed price per phase.

Controls

How the work is controlled

  1. 01

    Agents read far more than they write. We build the repository map and a written plan for each phase before any change.

  2. 02

    The agent that writes code never reviews it. Review agents have no shell access.

  3. 03

    Every change is a small pull request in your Git. Nothing merges without passing the harness and the review. A person approves each one.

  4. 04

    The AI bill is on the invoice. AI spend is billed at cost, on its own line, against a cap agreed before work starts. At the cap, work stops and we decide together.

  5. 05

    If moving the system won't pay for itself, the audit will tell you so plainly.

Industry evidence, not our work

14.5 weeks, primarily one developer

GitHub moved its Copilot runtime from TypeScript to Rust in about 14.5 weeks, primarily with a single developer. The port ran on a large test suite built before the agents started, and the team still traced regressions afterwards and fixed them. That is why our first deliverable is the test suite.

Our work · Mitsubishi Motors · client since 2014

Modernized on AS400, not replaced

Mitsubishi wanted an overhaul, not a full replacement. We built a web application on the existing DB2 database, integrated with the legacy RPG software and the company's security, and migrated the spreadsheet data the parts chain ran on.

Read the case

A date on your stack Microsoft ends support for .NET 8 and .NET 9 on November 10, 2026. An upgrade to .NET 10 is a Legacy Modernization with a deadline.

Plan the upgrade

Questions buyers ask

FAQ

What is agentic modernization?

Software delivery where one senior engineer directs AI coding agents against a codebase and owns the result, inside a set of tests that recorded the old system's behavior first.

How do you prevent regressions?

By building the tests before the migration, gating every phase on them, and keeping a regression window in scope.

Is AI-written code maintainable?

When it is reviewed, tested, and merged in small pieces, yes. Large unreviewed AI changes are a fair worry. The control is the process, not the model.

Does our code leave our network?

By default the work runs in your environment and your Git. It runs in ours only under a written agreement.

Which systems do you modernize?

Our record runs from IBM AS400, DB2, and RPG to current web, mobile, and cloud stacks. The audit says what the move takes for yours.