ShushuLab
SERVICES

Development itself,
and a process to keep it going.

We design each engagement around its requirements and risks: conventional system development and modification, or autonomous development line setup and maintenance.

Alongside web application and business system development and modification, we build autonomous development lines with project-specific design boundaries, working branches, tests and stop conditions. We establish development operations that can recover and continue after stops and interruptions, rather than letting AI write code without constraints.

SYSTEM DEVELOPMENT / MODIFICATION

System development and modification

From JPY 500,000

From improving existing systems to new web applications, business workflows, authentication and permissions, billing and external API integrations, CMS, operational tools, data flow and repository boundaries, test infrastructure and refactoring. We provide an individual quote after reviewing the system and requirements.

  • Web applications / Business systems
  • Modification and rehabilitation of existing systems
  • Authentication, authorization and role boundaries
  • Billing / Subscriptions / External APIs
  • CMS / Operational tools
  • Testing, regression prevention and maintainability
JPY 500,000 is the published starting price, not a ceiling. Individual quotes for larger, more difficult or higher-risk projects may be substantially higher.
AUTONOMOUS DEVELOPMENT LINE

Autonomous development line setup

From JPY 650,000 per line

We configure one project or repository as an independent development line that can repeat audits, implementation, testing, CI, recovery and resumption after interruptions. This is an ongoing operating process, rather than a single AI coding session.

  • Repository / Existing code audit
  • Dedicated working branch and prohibited operations
  • Design constitution / AGENTS.md / Worker mission
  • Autonomous worker + watchdog
  • Session continuation / Recovery after interruptions
  • Design for safely identifying and recovering worker-owned uncommitted changes after interruption
  • CI / Regression test integration
  • Stop conditions / Blocker reporting
  • Prevention of direct work on production / default branches
  • Initial trial operation and tuning
JPY 650,000 is the starting price for one line, not a ceiling. Additional lines, major prerequisite repairs, extensive redesign, new requirements and high-risk repositories require separate scopes and quotes.
NOT ONE-SHOT CODING

A single AI output
is different from a development operations line.

A typical single run

Human instruction
↓
AI run
↓
output
↓
stop
→

ShushuLab autonomous development line

Design boundary
↓
inspect → change → test → CI
↓
recover / restart
↓
continue

This diagram illustrates an operating concept. It does not guarantee code changes on every run or continuous operation without stopping. Necessary stops are treated as part of the safety mechanism.

PRICE / CONTINUATION

Maintenance by line, after setup.

SYSTEM DEVELOPMENTFrom JPY 500,000

Individually quoted based on requirements, scale and risk. Published prices are not ceilings.

LINE SETUPFrom JPY 650,000 per line

The starting price for initial setup, including the target repository, design boundaries, autonomous worker, monitoring, test integration and trial operation.

LINE MAINTENANCEFrom JPY 200,000/month per line

Maintenance is provided per independent line. Additional lines are a separate scope.

SETUP / MAINTENANCE SCOPE

Operations continue after setup,
including how the system can fail.

An autonomous development line involves more than installing a worker. The goal is maintainable development operations that account for interruption and resumption, CLI and model changes, and changes to repository structure.

INITIAL SETUP

What initial setup covers

  • Audit of existing code / Build / Tests / CI
  • Definition of working branches and prohibited operations
  • Documentation of architecture, invariants and permission boundaries
  • Worker mission / Watchdog / Restart paths
  • Design of session continuation, interrupted state, checkpoints and blocker reporting
  • Incorporation of safety boundaries into regression tests and CI
  • Initial trial operation and adjustments for the target repository
If the suitability assessment identifies a need to improve build reproducibility, tests or structure before broader autonomy, that remediation may be a separate scope.
ONGOING MAINTENANCE

What monthly maintenance can cover

  • Adapting to compatibility changes in workers / CLIs / Models / Tooling
  • Repairing broken CI and automation paths
  • Repairing watchdog, resume and restart failures
  • Adjusting the line when repository structure changes
  • Updating the design constitution and worker boundaries for approved product changes
  • Adding safeguards for newly observed failure modes
  • Periodic review and tuning of autonomous operations
Maintenance does not include unlimited new feature development, extensive redesign or production operations. New requirements and major changes are quoted individually.
INFRASTRUCTURE / REMOTE OPERATING MODEL

Dedicated PCs and servers
are normally provided by the customer.

In the standard autonomous development line model, ShushuLab builds the environment and line on a dedicated endpoint, server or PC provided by the customer, then maintains it remotely. The model does not assume ShushuLab owns or supplies the equipment free of charge.

Customer responsibilities

  • Prepare PCs or servers that meet the required / recommended specifications provided by ShushuLab
  • Assist with SSH / VPN / Remote access configuration as needed
  • Provide the permissions and connectivity needed for setup and maintenance
  • Own the equipment and handle its routine maintenance / replacement

ShushuLab responsibilities

  • Provide required / recommended specifications
  • Set up the development environment, autonomous worker, watchdog and related tools
  • Repository audit, test / CI integration and trial operation
  • Remote maintenance and line adjustments within the contracted scope
Remote support is standard. Hardware selection, procurement and on-site work are quoted separately. Travel, installation, shipping, equipment setup and on-site support where remote access cannot be provided are also scoped individually as needed.
SAFETY / TRUST

Define what can be delegated
before handing work to AI.

Autonomy is not an unlimited delegation of authority. Design boundaries, prohibited operations, working branches, tests and stop conditions are defined first, with failure, stopping and interruption treated as normal operating possibilities.

Branch isolation

Autonomous work is isolated on an approved working branch. The process does not make unauthorized direct changes to production or default branches.

Design constitution

Architecture, invariants, permission boundaries and prohibited operations are documented so that workers cannot reshape the design to suit themselves.

Tests as boundaries

Where possible, authorization, data integrity and important business conditions are encoded in regression tests and CI, expanding the boundaries that can be checked mechanically.

Fail closed

High-risk areas such as production changes, irreversible operations, money, legal matters and secrets are not silently delegated merely because an agent is running. Unclear high-risk decisions cause the process to stop and report.

Decisions retained by humans: Product direction, UX, behavior on actual devices, significant architectural changes, money, contracts, legal matters and production releases remain human decisions.
Work delegated to the line: Audits, repeated implementation, regression tests, CI, recovery, resumption and maintenance are automated and continued within documented boundaries.
PROJECT FIT

Projects that fit,
and projects that need preparation first.

A good fit

  • Repositories with reproducible builds and tests
  • Architecture and business rules that can be made explicit
  • Frequent recurring work such as maintenance, audits and refactoring
  • Important conditions that can be encoded in CI and regression tests
  • A desire to focus human time on direction and review

May need preparation first

  • No reproducible build
  • Almost no tests in high-risk areas
  • Requirements and permission boundaries are continually undefined
  • Production secrets or uncontrolled permissions mixed into the repository
  • A desire to automate even legal, financial or irreversible production decisions
ENGAGEMENT FLOW

From inquiry to the decision to go live.

Inquiry / Quote request
Review of the system, repository and requirements
Initial assessment of risk and suitability for autonomy
Agreement on the quote and scope of work
Contract
Initial audit
Environment setup / Construction and implementation
Trial operation and verification
Handover / Decision to go live
Remote maintenance as needed
SCOPE BOUNDARIES

Make the included scope
clear.

The autonomous development line setup price covers the agreed setup scope for one line. Additional lines are quoted separately.
Major new features, full redesign, new requirements and significant prerequisite repairs during setup or maintenance are not automatically included.
Production deployment, production operations and emergency response are not automatically included. Their scope is determined individually if needed.
Third-party costs for external APIs, models, licenses, infrastructure and similar services, as well as equipment costs, are reviewed per project.
ShushuLab does not, as standard, assume ownership of customer equipment or responsibility for its replacement or failure.
Published prices do not vary by sales channel. Actual contract amounts are quoted individually after reviewing the target, requirements, risks and scope of work.
SALES PARTNER

We are also recruiting sales partners to connect these services with the people who need them.

We seek independent business partners paid on results, who develop prospects and make initial contact, then hand over technical explanations, requirements, quotes, contracts and implementation to ShushuLab.

View partner conditions →
CONTACT

First, show us the system.

Whether you want conventional system development or an assessment of whether your existing repository can become an autonomous development line, we will review your target and goals and define an appropriate scope.

Only if the form is unavailable, contact [email protected].