Skip to content
SystemSolutions

Case Studies

Case Studies

From launching new services to business systems, AI automation, and cloud platforms. Client names are withheld; industry, challenge, what we built, and the result are shown as they are. Every project was led directly by our founder-engineer.

  1. New product launchAccounting firm group (CPA-led)2026

    Bookkeeping cloud product for an accounting firm group

    A bookkeeping cloud service offered as an in-house product by a CPA-led accounting firm group. We handle everything from requirements to design, implementation, AWS infrastructure, and operations, with continuous feature delivery.

    Challenge
    Bookkeeping work was spread across per-staff spreadsheets and manual steps; verification and re-entry grew with every client added.
    What we built
    Consolidated entry, review, and output into one cloud workflow designed around the firm's actual process, on an AWS foundation with IaC and CI/CD.
    Result
    In production as the firm's core workflow; the founder continues monthly improvements directly with the client.
    LaravelMySQLDockerAWSTerraform
    Our role: Ongoing contractor from requirements through development and operationsWeb & Business System Development
  2. AI & automationAccounting firm / social-insurance filings2026

    Auto-generating social-insurance filings from résumés and submitting via e-Gov

    Upload a résumé (PDF or image) and generative AI extracts the required fields to produce health-insurance and pension enrollment forms, then files them through the e-Gov electronic application API.

    Challenge
    Every new hire meant re-typing résumé data into forms, with transcription errors and back-and-forth checks.
    What we built
    Structured extraction with Gemini 2.5 Flash, automatic write-back into the official Excel template, live preview for review, then submission via the e-Gov API.
    Result
    Eliminated manual re-entry; from upload to filing data in minutes.
    LaravelLivewireGemini 2.5 FlashPhpSpreadsheete-Gov APIDocker
    Our role: Planning, design, development, and infrastructure end to endAI Agent Development & Generative AI
  3. New product launchVeterinary clinics2026

    Voice-to-chart transcription for veterinary clinics

    A web service that transcribes consultations and drafts the medical record with generative AI, so veterinarians spend less time on documentation.

    Challenge
    Post-consultation charting took too long and record quality varied by clinician.
    What we built
    Combined speech recognition with generative AI to produce a structured chart draft from the conversation; clinicians only review and confirm.
    Result
    Designed and built from scratch as a new service, through to the production environment.
    音声認識生成AI APIWeb アプリDockerAWS
    Our role: From concept through development and infrastructure for a new serviceAI Agent Development & Generative AI
  4. Integration & maintenanceHealthcare AI company2026

    Consolidating multiple databases and bringing operations in-house

    A fixed-scope project to consolidate databases spread across several systems and move vendor-dependent operations in-house, covering extraction, transformation, migration, and post-migration verification.

    Challenge
    Each system had its own database and every change required an outside vendor, inflating cost and lead time.
    What we built
    Designed the target schema, migrated in phases, and documented migration and operations so the client's own team can run it.
    Result
    Database operations moved from vendor dependence to the client's in-house team.
    データベース移行SQLAWSDocker
    Our role: Fixed-scope contract covering integration, migration, and enablementIT Consulting & DX
  5. Business systemBeverage wholesaler2026

    Daily sales-report system with AI-assisted visit history

    A business system where sales reps log customer visits, team leads review, aggregate, and approve, and administrators see the whole company — with visibility scoped by role, from entry to approval entirely on the web.

    Challenge
    Reports were scattered across paper and email; review and aggregation were slow and access control was ad hoc.
    What we built
    Three roles (rep, lead, admin) with distinct screens and permissions, a hierarchy-aware model that shows only one's own reports at any depth, and login by employee code to match the client's operations.
    Result
    Phase 1 released to production; entry, search, and approval now complete on the web.
    Web アプリ権限設計AWS
    Our role: Led by our founder as head of development at an AI startupWeb & Business System Development
  6. Cloud infrastructureMajor cram-school operator2026

    Production cloud platform for an automatic math-test generator, plus answer-sheet OCR evaluation

    Ran an automatic test-generation web system on a production stack of ECS Fargate (ARM), ALB, CloudFront, RDS, and ElastiCache, and evaluated generative-AI OCR accuracy on handwritten answer sheets.

    Challenge
    Needed a production-grade cloud architecture with reproducible deployment, and a realistic read on OCR accuracy before using it for grading support.
    What we built
    Documented deployment including ARM64 builds, health checks, and monitoring; digitized answer sheets, transcribed them with generative AI, and compared match rates across models.
    Result
    In production; OCR evaluation showed about 98% match on multiple-choice fields and informed a policy for handling free-text handwriting.
    AWS ECS FargateALBCloudFrontRDS (PostgreSQL)ElastiCache (Redis)DockerClaude
    Our role: Led by our founder as head of development at an AI startupCloud Infrastructure
  7. AI & automationProperty management company2026

    AI-assisted review of lease documents (proof of concept)

    A proof of concept where AI reads incoming lease documents and flags errors and cross-document mismatches, so reviewers only verify the flagged items.

    Challenge
    Three reviewers marked up documents by hand at roughly eight minutes per case.
    What we built
    Generative AI reads the contract PDFs, cross-checks fields, and overlays color-coded findings on the original documents.
    Result
    Demonstrated on real documents with a target of cutting review time from eight minutes to about one.
    生成AI（文書読み取り）PythonPDF処理
    Our role: From problem discovery through PoC developmentAI Agent Development & Generative AI
  8. Business systemReal estate / location data2026

    Location web app returning nearby facilities, school districts, and zoning from an address

    Enter an address to get nearby facilities with walking routes, elementary and junior-high school districts, and zoning with coverage and floor-area ratios — built on government open data and spatial database queries to avoid paid map-API costs.

    Challenge
    The incumbent commercial map API was expensive and its terms didn't fit the three required features.
    What we built
    After a feasibility study: geocoding via the national geospatial API, school districts and zoning via open datasets with MySQL spatial indexes, and facilities/routes via Google Maps Platform.
    Result
    Delivered all three features as a Laravel app and removed the dependency on a fixed-fee map API.
    Laravel 11MySQL SPATIAL国土数値情報国土地理院APIGoogle Maps Platform
    Our role: Feasibility study, design, and implementationWeb & Business System Development

How we work

Ship a new service on a ¥3M budget

We start with planning sessions, define an MVP around the one feature worth validating, and ship in 6–8 weeks. AI-driven development lets us deliver custom code at no-code prices and speed. No sales or PM layer — the founder-engineer works with you directly, so your budget goes into building.

Start with a planning sessionSee pricing guidance

Tell us about your challenge.

AI agents, cloud, or software — reach out by form or email. We work remotely with clients across Japan.

Initial consultation and quotes are free

Talk to usinfo@syssol.co.jp

Hamamatsucho Diamond Bldg. 2F, 2-2-15 Hamamatsucho, Minato-ku, Tokyo 105-0013, Japan