Who this is for
- Companies of 10 to 300 people who want to use AI and do not know where to start
- Back offices spending hours a day on invoices, purchase orders, data entry and report formatting
- Teams without a digital transformation department who need someone to own the project end to end
- Businesses that tried ChatGPT or a similar tool, saw a few people use it for a week, and watched it fade
The problem, stated plainly
Enterprise AI programs are priced and paced for enterprises. A mid-sized company does not need a transformation roadmap; it needs the three tasks that eat the most hours automated this quarter, and it needs the change to stick when the consultant leaves.
The failure mode we see most is adoption, not technology. A tool gets bought, a memo goes out, and nothing in the daily workflow changes. Airline operations have a name for the fix: standard operating procedure. If the new way of working is not written into how the job is done, it is not a new way of working.
What we build
Document processing with OCR, RPA and AI
Invoices, purchase orders, delivery notes and reports read automatically, validated against your rules and entered into your systems, with exceptions routed to a person. Keying errors disappear and the hours come back. We start with the one document type that costs you the most.
Internal AI assistants and chatbots
An assistant trained on your policies, manuals, contracts and FAQs, answering staff questions with the source cited, inside the tools your people already use. Response hours drop and answers become consistent; the assistant also tells you which questions your documentation does not answer.
Generative AI adoption that lasts
Tool selection, data and confidentiality rules, and then the part most programs skip: embedding the tool into specific workflows, writing the procedure, training the people and measuring use after 90 days. Claude, ChatGPT, Gemini or a private model, chosen for your data constraints rather than for fashion.
AI training and enablement
Half-day sessions from ¥30,000 for up to 30 people: strategy sessions for executives that end with a decision, hands-on workshops for frontline teams that end with a working prompt library. Delivered in English or Japanese, remote or in Tokyo.
Where the aviation model shows
Airlines run on procedures because the same job has to be done the same way by whoever is on shift. That is exactly what makes AI adoption stick in a small company: the new workflow is written down, the checklist includes the tool, and quality no longer depends on which employee happened to like it.
The mechanisms, and how they translate outside aviation, are in what aviation safety practice teaches other operations.
How an engagement runs
| Stage | What happens | What you get |
|---|---|---|
| Free consultation (30 min) | We hear how work flows today and where the hours go | A shortlist of two or three automatable tasks, ranked by return |
| Workflow improvement, implemented (from ¥50,000, about 2 weeks) | The top task automated on your real documents or questions | Measured hours saved and error rate |
| Whole-operation architecture, implemented (from ¥300,000) | Rollout across the process, integration, procedures written | A change your team owns and can explain |
| Training (half-day, from ¥30,000) | Executive or frontline sessions, customized | People who use the tools on Monday |
| Operate and improve (from ¥20,000 / month) | Monitoring, new document types, model updates | Automation that keeps up with the business |
Figures are indicative. Many SME engagements stay at the PoC-plus-training scale, and that is a perfectly good outcome.
What this will not do
AI does not fix a process nobody can describe. If the way an invoice gets approved differs by who is in the office, the first deliverable is the written procedure, and we will say so. We also do not recommend generative AI for tasks where a wrong answer is expensive and unreviewed; those get a human in the loop by design.