Work as usual.
Get AI agents built for your operations.
AI fails to deliver when it does not know how your company works.
Tsukumo turns PC operation logs into high-resolution data about your operations,
then goes beyond visualization to build AI agents that deliver results.
Initial consultation with a former McKinsey consultant
Why Tsukumo Labs
Start without burdening the frontline
Just install a lightweight client — there is nothing new for employees to learn or operate.
From actual operations, not memory
Even steps people perform without thinking are captured as data from real PC operation records.
Deployments that keep data inside
Choose from three options, including a fully closed configuration.
Problem
Business transformation depends on
how accurately work is understood
Traditional interview-led process analysis has structural limits that become costly at enterprise scale.
Interviews cover only what people can recall
BPR and operational improvement projects often spend months on manual interviews. The result depends heavily on the interviewer, and perceived workflows can diverge from what actually happens day to day.
Tacit knowledge is difficult to surface manually
High performers often rely on habits and judgment they do not consciously articulate. Interviews alone rarely explain why performance varies or how the best practices can be scaled across the organization.
AI initiatives fail when the underlying work is not understood
Without reliable process data, teams struggle to decide where AI should be applied. PoCs can miss the mark when they are built on partial or outdated assumptions about frontline work.
Consulting deliverables should reflect operational reality
Large process-mapping engagements can still end with interview-based overview reports. For enterprise teams, the question is whether the analysis captures enough ground truth to justify the investment.
How it works
Employees work as usual.
AI reads the workflow.
No upfront process mapping or heavy interview scheduling required.
AI analyzes PC operation logs and turns everyday work into process intelligence.
Capture
A lightweight application is designed to collect PC operation logs in the background, so teams can continue working without interrupting day-to-day operations. Employee notification and consent workflows are planned as standard.
Analyze
AI analyzes operation logs to identify process flows, person-dependent work patterns, and bottlenecks automatically.
Generate insights
The output is a business improvement report with a chat interface, designed to help prioritize opportunities using actual workflow data.
Build AI agents
Captured process knowledge feeds directly into requirements for Tsukumo Agent, a company-specific AI agent built to take over routine work.
Features
Why PC operation logs x AI
We use technology to reach the depth and speed of process understanding
that interview-led analysis structurally struggles to provide.
Process visibility with minimal frontline burden
No interviews firstNo interview scheduling, no note-taking, and no follow-up checks with the frontline. Process visibility no longer requires blocking your team's time.
Higher-resolution workflow understanding
Data-drivenBecause the analysis starts from PC operation data, it can surface time allocation, task sequences, and tool-switching patterns that are hard to reconstruct from memory.
Security architecture for enterprise review
On-prem readyThe product is being designed for on-premise and closed-network environments, with planned automatic masking for personal and confidential information plus employee notification and consent flows.
Beyond visualization to AI agents
Execution-focusedProcess knowledge from the analysis feeds directly into AI agent requirements, so engagements do not end with recommendations.
Use Cases
Where this approach is designed to help
From BPR and operational improvement to knowledge transfer and AI readiness,
Tsukumo Labs is built for teams that need a clearer view of how work actually happens.
Build an as-is view of work for transformation programs
Give executives and transformation teams a data-backed view of current operations early in a project. Instead of relying only on interviews, AI helps structure operational reality into a usable process database.
Turn high-performer know-how into organizational knowledge
Preparation routines before sales meetings, fast quoting workflows, and coordination habits held by experienced staff can be analyzed as behavior patterns and converted into repeatable operating practices.
Collect the operational knowledge AI agents need
Deciding where to apply AI, and what to hand over, requires data that describes how work is actually done. Workflows, decision criteria, and tacit know-how buried in employees' heads are turned into data from everyday work, building the foundation for AI adoption.
Beyond Visualization
Visualization is the starting point, not the goal
Once operations exist as data, finding issues, acting on them, and measuring impact become one continuous loop.
Even the impact of each change can be measured with the same operation data. The Tsukumo series supports this entire improvement cycle.
Understand current operations
Identify and prioritize issues
Execute improvements
Measure and improve again
Measurement feeds the next round of discovery, so the cycle keeps running as your business changes.
Deployment
Three deployment options for your security requirements
From a fully closed deployment that keeps every byte inside your network, to a SaaS model for the fastest start.
Structured so your IT security team can use it directly in security reviews.
Option A Closed environment | Option B Hybrid | Option C SaaS | |
|---|---|---|---|
| Overview | All components run inside your cloud environment. No data leaves your network. | Collection, storage, and the app stay in your environment. Only LLM processing uses our API platform. | All components are provided on our cloud for the fastest start. |
| External data transfer | None | Only masked analysis data is sent, encrypted (private connectivity available) | Yes (encrypted with TLS 1.2+) |
| LLM processing | Azure OpenAI Service / Amazon Bedrock in your tenant | Our managed platform (dedicated tenant) | Our managed platform |
| Lead time | 2-4 weeks | 1-3 weeks | 1-2 weeks |
Option A
Closed environment
- Overview
- All components run inside your cloud environment. No data leaves your network.
- External data transfer
- None
- LLM processing
- Azure OpenAI Service / Amazon Bedrock in your tenant
- Lead time
- 2-4 weeks
Option B
Hybrid
- Overview
- Collection, storage, and the app stay in your environment. Only LLM processing uses our API platform.
- External data transfer
- Only masked analysis data is sent, encrypted (private connectivity available)
- LLM processing
- Our managed platform (dedicated tenant)
- Lead time
- 1-3 weeks
Option C
SaaS
- Overview
- All components are provided on our cloud for the fastest start.
- External data transfer
- Yes (encrypted with TLS 1.2+)
- LLM processing
- Our managed platform
- Lead time
- 1-2 weeks
* Other cloud platforms can also be supported. Please contact us for details.
Security & Privacy
Designed around security
and employee privacy
Because PC operation logs can be sensitive, security and privacy are treated as core product design requirements rather than afterthoughts. The items below reflect the security principles being built into the product.
Tsukumo Labs is intended to understand business processes, not to monitor individuals. Collected data is designed to be analyzed after aggregation and anonymization, and we do not provide externally identifiable personal activity data.
We are developing the product with APPI-aware design and operations in mind, and we can support enterprise security reviews during PoC discussions.
On-premise / closed-network deployment
Designed to support deployments that remain inside the customer network without sending data externally.
Automatic masking of sensitive information
Planned AI-assisted detection and removal of passwords, personal names, and confidential information before analysis.
Employee notification and consent flows
Planned workflows for purpose notification and opt-out support aligned with APPI-aware operations.
Access control and audit logs
Designed with role-based access control and audit trails for administrative and analysis actions.
Retention controls and deletion
Planned controls to automatically and irreversibly delete data after customer-defined retention periods.
About Us
Company Overview
| Company | Tsukumo Labs, Inc. |
| Founded | February 2026 |
| Address | Shibuya Dogenzaka Tokyu Building 2F-C 1-10-8 Dogenzaka, Shibuya-ku, Tokyo |
| Capital | JPY 2.5 million |
| Business | Generative AI product development Generative AI adoption support and consulting |
Sumiki Hori
CEO / Founder
After graduating from the University of Tokyo, Sumiki worked at McKinsey & Company on operational transformation and new business strategy across manufacturing and TMT clients in Japan, the US, and Europe, and was promoted to manager in under four years. He later contributed to core business growth at Timee around its IPO period before founding Tsukumo Labs with a focus on AI and Japan's productivity challenges.
FAQ
Frequently Asked Questions
How do you protect employee privacy?
Tsukumo Labs is designed for process-level analysis, not individual surveillance. Collected data is intended to be anonymized and aggregated before analysis. Employee notification and consent workflows are planned as part of standard deployment, with APPI-aware operation as a design premise.
How long does implementation take?
Environment setup typically takes one to four weeks depending on the deployment option. Time to the initial report depends on the scope of target operations, and we share a concrete plan during the initial consultation.
How is pricing handled?
We currently discuss PoC scope and pricing based on the size of the target organization and security requirements. Please contact us for a tailored discussion.
Where is data stored?
It depends on the deployment option. In the closed and hybrid options, operation log data stays inside your environment. Details for the SaaS option are discussed based on your security requirements.
Does Tsukumo Labs require integration with existing systems?
No. Tsukumo Log records PC operations directly, so data collection requires no changes or API integrations on the target systems, and it captures work that spans SaaS, internal, and legacy systems.
How is this different from consulting?
Traditional BPR consulting often relies on interviews and can take months before a report is produced. Tsukumo Labs is designed to reduce the data collection and analysis burden by using AI to analyze PC operation logs. It does not replace consulting judgment; it strengthens the discovery and analysis phase with operational data.
Start by turning your work
into process data.
In an initial discussion, we will review where your workflows may have room for improvement
and whether a PoC is the right next step.
A former McKinsey consultant will help structure your operational questions.
Short inquiry form