ACLabGet in touch

ACLab / Public portfolio

Try the workflow.
Explore the design.

I’m Tomoya Nishimura. These independently built demos show how I turn a workflow into an application. Each project includes the problem, an implementation decision and a working demo. They are not customer case studies.

Explore the demos

Featured project / Inventory teams / Fictional data

StockFlow.
Approve the request. Then move the stock.

For teams sharing inventory, StockFlow demonstrates how to request a stock movement, check the proposed quantity and approve it before updating the inventory. The request, decision and activity history stay together.

Try StockFlow
StockFlow shows fictional USB-C cable inventory, an approved equipment request and its activity history.
Actual StockFlow demo screen with fictional data. The recorded approval shows stock changing from 48 to 45 units; the form previews a separate next request. Open the image to inspect it at full size.

Example: equipment for a workshop

  1. RequestA staff member requests USB-C cables for a workshop. Submitting the request leaves stock unchanged.
  2. ReviewA reviewer checks the item, quantity and reason before deciding.
  3. ApproveApproval updates the inventory and records the movement in the activity history.

You can play both roles in this demo. The example illustrates the workflow; it is not a customer case study.

The implementation question
How can a workflow prevent a repeated request or an outdated approval from changing stock incorrectly?
Built with
Cloudflare Workers and D1 store the demo workflow in a cookie-isolated sandbox. Repeated requests are handled once; approval rejects an outdated stock version and records an audit history.
Demo limits
Fictional stock and demonstration reviewer actions, not authenticated production roles. There is no AI inference or connection to a live warehouse. Production use requires identity, access controls and operational review.

Browser AI / Object vocabulary / English UI

Word Lens.
Objects become vocabulary.

For language learners who want to connect vocabulary to everyday objects. Try a sample photo, select a detected object and explore its name and pronunciation in another language before opening your camera.

Try Word Lens
Word Lens detects a chair in the sample photo and displays the Spanish word silla.
Actual Word Lens sample result. Photo: Chair by Antonio Borrillo, CC0. The screenshot shows sample-photo inference, not a live camera capture.
The implementation question
Can someone inspect a real result before deciding whether to grant camera access?
Built with
Object detection runs in the browser. Static assets are deployed with Cloudflare Workers. Vocabulary uses authored translations; pronunciation uses available device voices.
Demo limits
Only supported object categories are detected, and recognition can be wrong. Images stay in the browser. Some device voices may use an online speech service.

Data integration / English UI / Local file processing

CSV Bridge.
Check the data before import.

For operations teams preparing CSV imports. Map columns, inspect invalid rows and download a normalized file before taking it into another system. Start with the fictional sample or process a local file in your browser.

Try CSV Bridge
  1. LoadChoose a CSV or start with the included fictional sample.
  2. ValidateMap columns and inspect invalid or duplicate records.
  3. ExportDownload checked rows with formula-like text prefixed; review spreadsheet import settings.
Interactive portfolio demo. The sequence describes the app; it is not a customer system screenshot.
The implementation question
How can an import surface invalid data before it reaches another system?
Built with
CSV parsing, mapping and validation run in the browser. No file upload, database or third-party service is involved.
Demo limits
This is a bounded CSV preparation demo, not a live connector. Review the destination schema before importing. Inputs stay in memory until you close or reload the page.

Implementation planning / English UI / Rule-based output

Japan Pilot Desk.
Define the pilot before the build.

For AI and SaaS teams planning a Japanese customer pilot. Turn a fictional workflow and selected requirements into draft acceptance criteria, responsibilities and open questions to review with the customer.

Try Japan Pilot Desk
  1. DescribeDescribe a workflow and select its operating requirements.
  2. ReviewInspect the resulting delivery steps and unresolved questions.
  3. ExportDownload an implementation brief for review.
Interactive portfolio demo. The sequence describes the app; it is not a customer system screenshot.
The implementation question
What needs to be agreed before a Japanese customer pilot begins?
Built with
A deterministic rules engine turns selected requirements into a browser-only implementation brief. Input stays in browser memory.
Demo limits
No AI generation, customer submission, legal advice or compliance assessment. The output is a starting point that needs review with the customer.

Build a workflow around your business.

The initial consultation is free, and you do not need a finished specification. Have an inventory, approval or data-preparation process to improve? Tell me how it works today, who needs to review changes and which tools you use. I can help define a pilot, build the workflow and document how to operate it. I also support global AI and SaaS teams implementing their products for Japanese customers.

Discuss your workflow with ACLabJapan Launch & Implementation