Your files stay local · An AI colleague that gets work done
Tell Otoha what you need. She gets it done.
Otoha researches, writes documents, and builds spreadsheets and slides in your project folder. Progress stays visible, and you approve key actions.
macOS 13 or later · Apple silicon · Open source

Hand over a task in three steps
No prompt engineering or complex setup. Delegate work the way you would to a colleague.
- 1
Create a project
Pick a local or shared folder and drop in your materials. Every project keeps its own context.
- 2
State the goal
Describe what you need in plain language, such as “Summarize last month’s sales and build an 8-slide deck.”
- 3
Review and collect results
Follow progress in real time. Completed Word, Excel and PowerPoint files are saved straight to Results.
Core capabilities
Research, document creation, file organization and recurring checks in one workspace.
Project-based workspace
Each project has its own folder, board and deliverables so progress is clear at a glance.
Built-in web research
Searches and reads pages in the built-in browser, showing you the exact sources inspected.
Office documents and slide decks
Creates Word, Excel, PowerPoint and web slide decks directly inside your project folder.
Scheduled recurring runs
Run weekly or monthly aggregations, reports and ledger checks automatically on a set schedule.
Version history and approval
Saves a snapshot before editing files and asks before running external tools, so changes are easy to undo.
Files stay on your machine
Accesses only the folders you allow. With a local or in-house model, nothing ever leaves your network.
For teams & enterprise
Solve the six workflows that drain the most hours
Companies often budget tens of thousands of dollars for custom systems just to fix everyday bottlenecks. Otoha starts with the files you already have, and our engineers add custom skills for your workflow.
Scattered documents and strict confidentiality
Manuals, specs and past proposals sit across personal and shared drives, and sensitive files cannot be uploaded to public AI.
Indexes shared folders locally and cites the exact file, page and section in every answer. Pair with a local model so data never leaves the machine.
Repeated inquiries and know-how stuck with veterans
Teams spend hours every month answering the same client and internal questions while terminology drifts across staff.
Drafts cited answers from past tickets and technical docs, searching across English, Japanese and Chinese sources.
Monthly reports built by hand in Excel
Sales data is exported to CSV and combined by hand with cost figures, while inventory orders rely on gut feel.
Turns exported CSV/Excel files into totals, charts and reports on a monthly schedule, and suggests order quantities from historical trends.
Quote and sample requests retyped from emails
A lean sales team manually copies dealer requests from emails and attachments into spreadsheets and core systems.
Extracts required fields from emails and attachments into your quote template or import-ready CSV, asking before anything is sent.
Excel ledgers only one person understands
Hundreds of equipment and contract renewal dates live in complex spreadsheets that break when staff rotate.
Scans the ledger on a weekly or monthly schedule, flags upcoming renewals, and standardizes the handover workflow.
Legacy core systems with no documentation
Decades-old in-house systems are expensive to modify or export data from because nobody knows the full logic.
Reads legacy source code and old specs to document processing flows and data fields before vendor selection and migration.
Forward deployed engineers on site until it sticks
Software alone rarely changes how a team works. Our FDEs join you on site to map bottlenecks, build custom AI skills, and train your staff.
- 1
Map the work
Identify which tasks consume the most hours and pick the highest-return starting point.
- 2
Pilot in one team
Verify results on real documents and data within a single department.
- 3
Build AI skills
Package your templates, pricing rules and system formats into reusable Otoha skills.
- 4
Train and hand over
Train users, roll out across the company, and transition maintenance to your team.
Tell us which workflow takes the most time
No formal requirements document needed. Share how the job is done today and we will propose a pilot plan.
How Otoha compares
Chat AI is quick to start but stops at conversation; custom systems and workflow builders fit your rules but take months to build. Otoha combines both: local files, flexible models, central IT policy and custom skills.
| Otoha AI | Chat AI (ChatGPT, Claude) | Custom systems & workflows (Dify) | |
|---|---|---|---|
| Models | Choose GPT, Claude, Gemini, DeepSeek, or connect in-house and local models | Single provider only | Multiple models, configured by developers |
| Where files live | Stay in local or shared company folders; no upload required | Uploaded to external cloud servers | Imported into self-hosted servers you maintain |
| Keeping data in-house | Nothing leaves the company with local/in-house models; IT can restrict allowed models | Processed on external cloud servers | Possible if self-hosted and operated in-house |
| Central IT policy | Enforce allowed models, permissions and confirmations via MDM or policy file | Admin console on enterprise plans | Must be custom-built |
| Workflow fit | Add custom AI skills per workflow with FDE implementation support | General-purpose chat | Every workflow designed and coded from scratch |
| Actual execution | Creates Office files and slides, organizes folders, and keeps version snapshots | Returns text replies and drafts in chat | Runs within fixed pre-built pipelines |
| Time to launch | Install and start right away; pilot in one team before scaling | Immediate after sign-up | Months of scoping and development |
Based on typical usage as of October 2026; check each product’s official plans for specifics.
Supported models & smart routing
Connect your own OpenAI, Anthropic, Google or DeepSeek account, or run local models on your machine.
- GPTOpenAI
- ClaudeAnthropic
- GeminiGoogle
- DeepSeekDeepSeek
- Local modelsFreeRun open models on your computer via Ollama or LM Studio at zero cost, keeping every byte on the device.
Today, add your own API keys in Settings (billed by each provider). Once Otoha's cloud service opens, you can use all cloud models with a single monthly plan.
Pricing
Free to use today with your own API keys or local models. Otoha's cloud service is coming soon with monthly credit and zero key management.
- TodayUS$0
- Use your own provider API keys
- Connect local models at no charge
- PlusUS$20/ monthEveryday work
- Monthly cloud credit allowance
- All supported cloud models
- Automatic model routing
- ProUS$50/ monthHeavy daily workloads
- Monthly cloud credit allowance
- All supported cloud models
- Automatic model routing
- MaxUS$150/ monthIntensive research, writing & decks
- Monthly cloud credit allowance
- All supported cloud models
- Automatic model routing
Credit pack US$50
Top up anytime. Credit packs roll over month to month and never expire on a monthly reset.
Enterprise deployment & FDE support (Custom quote)
On-site workflow mapping, custom AI skill development, internal model integration, MDM policy rollout and team training.
- Plus, Pro and Max show planned pricing for the upcoming cloud service; final terms will appear at checkout.
- Plan credit refreshes every 30 days without rollover and is consumed before credit packs. Payments are processed via Stripe.
Frequently asked questions
- Do I need to pay Otoha right now?
- No. Enter your own OpenAI, Anthropic, Google (Gemini) or DeepSeek API key in Settings and pay the provider directly, or run a local model for free.
- When will Windows, Linux and mobile apps be available?
- The macOS app is available today. Windows and Linux desktop builds, along with iOS and Android companion apps for remote control, are in active development and coming soon.
- Are my company files uploaded to external servers?
- Files stay in your local folders or company file server. Cloud models receive only the excerpts needed for the active task; with local or in-house models, nothing leaves your network.
- How do I connect a local model?
- Install Ollama or LM Studio, download a model, and add its local address under Local service in Otoha Settings.
About us
Otoha AI Inc. is a Tokyo-based AI company researching AI theory, reinforcement learning and general-purpose AI agents, with papers published at international conferences. We turn frontier research into dependable everyday tools.
- Company
- Otoha AI Inc.
- Location
- Tokyo, Japan
- Business
- AI theory, reinforcement learning and general-purpose agent R&D; AI product development and enterprise deployment
- Contact
- support@otohaai.com
Products
- Otoha AI
An AI colleague that completes real work while keeping your documents in-house.
- Otoha AI Player
A music player with synced lyrics and subtitles where you can ask AI about any line (iPhone, Mac).
- FormAgent
Turns your website into an AI assistant that guides visitors and lifts conversion.
- Ask Sun Tzu
Applies The Art of War to workplace decisions and gives one concrete action for today (iPhone, Android).
Partner programme
We partner with agencies and system integrators to bring Otoha into enterprise workflows.
- Lead customer proposals, deployment and training with technical backing from our engineers.
- Provide your own managed model endpoint as the customer’s trusted AI provider, locked in via admin policy.
- Build domain-specific AI skills tailored to your clients’ industries.
Get started with Otoha
Download the app, connect your model account or a local model in Settings, and hand over your first task.
macOS 13 or later · Apple silicon · Open source
Windows · Linux · iOS · Android coming soon