Find the question quickly
Summarize the symptoms and context, while spotting missing details that a person needs to ask about.
04 / Software & AI · Planned integrations
Less copying
More context
Useful AI starts with a specific task. For us, that means finding the missing detail in an enquiry, preparing a reply and keeping a person responsible for the answer.
Walk through an exampleWe use AI to support development. These Claude integrations are not live yet.
A hand-written sample
A useful draft starts by knowing
what we don't know.
“My tablet won't charge.
What should I check?”
A useful draft starts by knowing
what we don't know.
Model missingStaff review first
Product tour · No login needed
The task
Then the assistance
Two views of the same idea: a repair workspace that keeps context, and a planned assistant that helps staff read it.
Repair ERP · In developmentThe check belongs
The check belongs
with the device
Inspection information sits with the work order instead of being copied into another conversation. The public demo lets you follow a fictional request and its related records.
Explore the sample workspace ↗Fictional data only. A separate read-only preview, not internal operations.
Planned Claude workflowA draft
A draft
Not an automatic reply
The enquiry becomes a short brief: what is known, what is missing and what a member of staff should check before responding.
This hand-written sample does not call Claude or send a message. It is not a live product feature.
Illustrative enquiry brief
Tablet charging enquiry
- Known
- The tablet is not charging.
- Missing
- Device model, charger used and whether the port shows damage.
- Draft for staff review
- “Could you share the tablet model and the charger you're using? Please don't force the connector if the port looks damaged.”
Staff review before any response is sent.
Where Claude fits
Small tasks
Clear responsibilities
We're developing the workflow before treating an AI output as a product feature. A successful pilot needs useful answers, sensible costs and an understandable review process.
Give staff a starting point
Prepare a draft from approved context for staff to review. No automatic sending or repair decision.
Help with the building
AI supports development, test cases and documentation. Product behaviour still needs code review and testing.
Test before connecting
Use synthetic enquiries to examine quality, unsupported claims, privacy boundaries and API cost before a customer-facing integration.
See the underlying projects
Not just a description
Explore the original repair-table and inspection components, rendered locally with fictional data and no live customer records. Or look at a static capture of the monitoring workspace.
Let's talk
Start with the problem
We'll take it from there
Questions about the projects, device enquiries or a software collaboration? Email us or say hello on KakaoTalk. Our services are still in development.