01
Compounding knowledge
Go beyond storing information. We are designing ways to reuse reviewed results, decision rationale, and lessons in the next question. This cycle of reused experience is what we call compounding knowledge.
JARVIS · AI OS FOR AX
In development
Today’s knowledge. Tomorrow’s progress.
We are building an AI work environment that turns knowledge into ideas, connects work with appropriate AI, and carries reviewed experience into the next task.
- 01
Knowledge
Find context in information and experience.
- 02
Ideas
Connect knowledge to frame the next question.
- 03
Execution
Develop drafts and work solutions with AI.
- 04
Reviewed experience
Check evidence and outcomes, then capture what was learned.
Reviewed experience returns as knowledge — so the next task need not start from scratch.
THREE DIRECTIONS WE ARE DESIGNING
Knowledge becomes an asset when it contributes to the next task.
JARVIS is being developed to connect scattered knowledge, work with AI, and human review in one continuous flow.
02
AI matched to the work
Research, writing, and review serve different purposes. We are developing a router that considers the purpose of a task and the nature of its information when connecting work with AI.
03
Reviewable execution
Develop ideas into drafts and action plans with evidence and open questions to inspect. People make important judgments, and reviewed outcomes inform the next task.
Individual knowledge workers
Research and proposals
Let the last research project inform the next proposal.
Consider starting a new proposal: finding the same sources again and struggling to recall the reasoning behind earlier notes.
Use scenario under design
- Find context relevant to the current question in research and notes.
- Connect evidence to develop the argument and draft of a proposal.
- Record verified evidence and reasons for revision to inform the next research task.
What carries forward: reviewed evidence, developed ideas, and lessons from revision.
Public institutions and businesses
Public institutions
Keep the reasoning when responsibilities change.
Policy, project, and work records accumulate, while the next person must reconstruct the reasons behind decisions and the remaining tasks.
Use scenario under design
- Gather work history and evidence from information cleared for its intended use.
- Draft a handover with decision rationale and unresolved questions.
- Carry records reviewed by the responsible person into the next task.
What carries forward: reviewed evidence, decision context, and follow-up tasks.
Businesses
Bring different colleagues’ project experience into the team’s next proposal.
Project materials and review records are scattered across colleagues, making it difficult for a team starting new work to revisit earlier decisions and lessons together.
Use scenario under design
- Gather relevant context from different colleagues’ project records cleared for their intended use.
- Review evidence and decision rationale as a team while developing action plans and proposal drafts for the next project.
- Record outcomes and reasons for revision reviewed by the team for reuse in the next project or proposal.
What carries forward: team-reviewed project knowledge, action plans, and lessons for the next task.
THE CONNECTION WE ARE WORKING TOWARD
JARVIS as a shared foundation. Concrete work as the experience.
The direction is to apply connected knowledge and reviewed experience to core work applications such as writing and handovers. Each product has its own preparation stage; they are not currently offered as one integrated application.
JARVIS · Connected knowledge / Task-aware AI / Human review
Insu-e · Handovers
A product direction for carrying decision context and open tasks to the next responsible person
View this product’s direction →Logical Writing · Documents and thinking
A product direction for shaping the structure of ideas and expressing them in writing
View this product’s direction →DESIGN PRINCIPLES
Work with AI. Keep people responsible for judgment.
Human judgment
We design workflows in which people own important decisions and approvals.
Evidence to inspect
Our principle is to examine sources and open questions alongside the result.
Scope of information use
Our principle is to establish permitted use according to the work and the nature of the information.
CURRENT PUBLIC SCOPE
We are developing the core product direction.
Knowledge cycles and task-aware AI connections are under development and validation. The flow and scenarios on this page describe the design direction. No public product access or confirmed launch schedule is available yet.