Start
Start with real needs
A brief, a task, and an idea first become something that can be handled by the workbench.
The operating layer for AI work
Finish AI workand keep it in one place.
Raydo is a local-first system for AI work. Runs, artifacts, approvals, cost, and provenance live in one workspace, so the next job can pick up where the last one left off.
Local-first · macOS · Windows · Linux · Codex · Claude Code · Gemini CLI
raydo.app · Workspace
From requirement to completion
A request comes in. A deliverable goes out.
Raydo lets non-technical users follow each step without needing to decode the machinery behind it.
In progress
Keep the work going
Progress, instructions, documents, and people involved all follow the same work.
Check
Check before handing it over
The person who should confirm can look at the results first and then decide whether to enter delivery.
Ready
Give clear delivery
The final result can be previewed, downloaded, shared, and used next time.
The flow stays simple: start the work, move it forward, review the result, and share the delivery.
Four access layers
Models, adapters, connections, and channels in one cloud.
A compact logo cloud that shows the access surface Raydo brings together.
LLM
Adapter
App Connection
Channel
Why Raydo exists
AI keeps getting stronger. The work around it still falls apart.
There are more models, more tools, and more entry points than ever. Starting is easy. Continuity is hard. Reuse is rare.
Tools remain fragmented
Chat, writing, code, assets, and distribution channels all live in different windows, so real work gets split into fragments.
Work loses continuity
Switch the model, window, or device and the context resets. Every run starts from zero again.
Outcomes do not come back
Getting one answer is easy. Leaving behind something reusable, traceable, and worth building on is much harder.
Raydo's approach
Finish. Gather. Reuse.
All three have to be true at once. Most AI tools only do one well. Raydo is built to do all three.
1
Finish
Work has a process, constraints, approvals, outputs, and an actual endpoint. It is not just one more generation.
2
Gather
Runs, outputs, versions, cost, approvals, and provenance stay together by default, so nothing has to be hunted down later.
3
Reuse
The next job can start from the last one, with roles, prompts, seeds, voice settings, and boards already in place.
The loop that matters
Work done should not disappear.
Raydo treats every result as something you can keep, understand, and build on the next time around.
Done
Complete
The mission has a real end point, not just another answer in the chat.
Keep
Gather
Results, review notes, documents and context stay together.
Reuse
Reuse
The next work can continue from the verified results.
Inside Raydo
One workspace, five places where the work actually happens.
Dashboard shows what matters now. Inbox moves work forward. Role Pack sets up the team. Connections wire in tools. Delivery shows what is ready to hand off.
Raydo · Workspace
organizationroleInbox
Current work, role networks, entry points, recent deliverables, approvals, and inbox items all surface on the first screen.
One workspace, fewer tabs
Work flows within Raydo, no longer spread across a bunch of tabs.
Each interface is only responsible for one clear thing, and users follow the work instead of chasing the window.
Start
Dashboard
Let’s first look at what needs to be dealt with today.
Advance
Inbox and Kanban
Push the work into a clear state.
Prepare
Role Pack
Use prepared roles so you do not have to rebuild the team every time.
Connect
Connections
Connect tools and channels in one place.
Delivery
Delivery
Put the results away and give them to others to review and confirm.
Platform capabilities
It is a system, not another chat shell.
Eight capability layers, one operating system. Each appears where it is actually needed instead of being dumped onto the homepage all at once.
Organization and roles
Companies, projects, templates, skills, and workflows shape how a real team operates when work is actually running.
Chat execution layer
Unify sessions, roles, contexts, tool events, and permission confirmations in every runtime.
Runtime connections
Codex, Claude Code, Gemini CLI, Hermes Agent, and ACP runtimes are all connected through the same desktop.
Models and Providers
OpenAI/Claude/Gemini/DeepSeek/Qwen/Moonshot/Doubao/Zhipu/Ollama/OpenRouter.
Channels and Connectors
Notion, Feishu, Slack, WeChat, DingTalk, Telegram, Discord, and Google Workspace all come back into the same operating layer.
Media and Assets Workbench
Images / Videos / Audio, with Media Bin, Source Monitor, TTS / ASR, and shared asset library.
Operations, approval, and traceability
Runs / Approvals / Scheduled Tasks / Live Artifact / Project Delivery can all be retained and audited.
Local-first vs. portable
Tauri + Rust + SQLite. Carrying case, runtime payload, offline and USB drive delivery are all ready.
Review and Delivery
The results should become something that others can trust.
Raydo makes delivery legible: preview, review, confirm, share, and retrieve it later.
Preview
Preview results
No need to go through the chat history, just open the latest results directly.
Check
Check for changes
Before accepting, look at the key context surrounding the outcome.
Confirm
Confirm delivery
The review steps for the team or client stay on the same page.
Delivery
Share or download
Send a clear delivery page, or just take the documents away.
Delivery chain
From a run to a handoff, everything stays connected.
Raydo keeps the result, review steps, and delivery context together so people can understand what happened and trust what is being sent.
Work starts with a run
Progress, notes, and approvals stay attached to the same work.
The result becomes a Live Artifact
Preview, refresh, restore, export, and share without losing context.
Delivery is ready to review
Teams or clients get one clean place to inspect and accept the outcome.
Workflow / Live Artifact / Project Delivery
Three surfaces, one workspace
Desktop for the work, CLI for diagnostics, read-only web for remote review.
Raydo shows up where the work actually happens: on the desktop, in the terminal, or in a browser for remote review.
Desktop
Main entry point for organization, roles, capabilities, processes, execution, approval and delivery.
macOS · Windows · Linux
Tauri + Rust + SQLite, local first by default
Supports portable bag and USB flash drive, offline delivery
raydo CLI
Diagnostics, scripting, and automation access — another view into the same workspace.
doctor · status · logs · config
skills · workflow · ops · opc
provider-auth · scheduled-tasks · update
Read-only web
Powered by manager_api, with front-end and back-end guards that block writes by design, making it safer than handing over remote control.
Inspection, review, acceptance
You can take a look at it without a desktop
Clear borders, not a fake console
Work in the right place
Do the work locally. Let others review it safely.
Raydo keeps the main workflow on the desktop and gives teammates or clients a safer way to review it without handing over control.
Execute
Desktop doing things
The main workbench is closest to your files and daily tools.
Review
Check when needed
Diagnostics and inspections can exist but do not take away the main thread from ordinary users.
Review
Share view link
Others can view and accept it, but will not gain write permission.
Real use cases
Raydo is built for three kinds of work.
Not a demo workflow — the product is designed for these operating patterns. Start with the one closest to yours.
1
AI production
Script → Character → Storyboard → Generate → Retry → Cost → Approval → Delivery, the whole process is in one running state, suitable for short drama studios, advertising agencies, and MCNs.
2
Research reports and analysis
Chapters, citations, traceability, approval, export - leave it once and you can continue directly next quarter. Suitable for research reports, competitive product analysis, and consulting reports.
3
Implementation and operation
Desktop work, CLI diagnostics, read-only web acceptance - with a portable package, it can be used for USB flash drive delivery and offline distribution.
What keeps it grounded
Local-first by default. Critical steps stay reviewable.
Capabilities are expanding and boundaries remain clear.
Local-first
The workbench lives on your machine by default.
Remote read-only
The web portal is read-only by default, and write operations are guarded.
Human in the loop
Approval nodes are first class citizens, not side panels.
Built-in provenance
Who did it, what changed, when it happened, what it cost, which version ran, and what came out all stay attached.
Portable delivery
USB flash drives, offline payloads, and enterprise distribution paths are all in place.
Before you install
Six practical questions to answer before you install.
The short version is enough to tell you what Raydo is and where it fits.
Start here
Download Raydo and judge it by the work it leaves behind.
Runs on macOS, Windows, and Linux, with support for portable packages and offline delivery.
macOS · Windows · Linux
Portable package support for USB handoff and offline distribution.
Local first, suitable for teams
Role Packs, approvals, and project delivery are already built into the product.
Documentation and Contact
Need deployment support or a trial? Contact us directly.