For non-coders, founders and professionals
The same Claude Code developers swear by, with buttons instead of commands. I run my agency on it. My wife, who has never opened a terminal, runs her Instagram on it.
+ a community for you to discuss, learn and share.
Free plan, no card. Windows 10 and 11. You need a Claude subscription (Pro or Max).
Everyone told you AI would change everything. Then you opened Claude Code and met a black window with a blinking cursor. The developers got a superpower. You got a tutorial.
OrcStrator is the part in between. Same AI the developers use, on a board you can actually read.
Start small. Then keep handing over the stuff you hate doing.
New headline, swap a photo, fix the broken form, add the cookie banner your lawyer keeps mentioning. Ask in plain words, see the result, say yes.
Every job is a card on a board: Ready, Running, Review, Done. Several agents work at once (three on the free plan), and the ones that need you turn amber and say why.
Edit a recording into a long video plus shorts. Write the week's posts. Turn a call into a case study. Put it on a schedule and it happens without you.
Build and deploy the internal system your business runs on, and share it with your employees. The thing an agency would quote you a fortune and a quarter for.
I set up a fresh copy of OrcStrator for a made-up bakery and took real screenshots. Click any picture to see it full size.
Everything your agents did today, what needs you, and how much of your limit is left, live on one screen. Every routine run of the week sits below it, with how long it took and what it cost.
Dashboard
All your Claude Code chats side by side on one screen. When Claude says "run this command", one click runs it for you.
A beautiful interface for Claude Code
Cards from Backlog to Done, every project in the sidebar, plus List and Calendar views. Start a card and an agent picks it up.
Pipeline board, views and project sidebar
Each chat works on its own copy and knows what the other agents are in the middle of, so two of them never rewrite each other. Here one agent leaves the page alone because another one is still editing it.
Unsaved changes and Finish
Drop or paste screenshots into a chat, and open any picture full size.
Images: drag and drop, paste, viewer
Sign in to Claude inside the app and switch accounts, no terminal.
Account switcher
Switch tools, skills and plugins on and off per agent, so each chat loads only what the job needs and your limit goes further.
Tool, skill and plugin management
Finished today
While you're on calls, the cards keep moving. This is what a normal day looks like for the made-up bakery.
Your chats, files and projects never leave your machine. Your work goes straight from your computer to Claude, on your own Claude account. Nothing passes through me: no proxy, no resold usage, no middleman.
You sign in once and you're done. The only thing I ever see is your email and which plan you're on.
The one catch: your computer has to be on for your agents to work.
Developers spend weeks tuning this stuff. OrcStrator does it in the background, so you can keep not knowing what it means.
Prompt caching
Chats are kept warm for an hour, so your next message doesn't make Claude reread the whole conversation from scratch. Faster replies, less of your limit burned.
Compaction
Big chats are shrunk into a tight summary automatically, and quiet ones are compacted just before Claude would forget them. The next message starts fast and cheap.
Session resume
Park a job today, press Start next week, and the same conversation carries on with everything it knew. A long goal keeps going, turn after turn, until it's done.
Secrets and API keys
By default, Claude Code saves your whole session to disk, including every password and API key you drop in the chat. When you close a chat, OrcStrator scans the full transcript (no tokens burned, it never asks Claude) and deletes everything it can identify as a password, an API key or anything else that shouldn't be there.
Rate limits
Stop keeping a second monitor open to refresh your Claude usage page. Your limits sit on screen in real time, with a plain heads-up like "Out of quota in 2h 13m" while there's still time to do something about it.
Subagents and tool loading
Searching and reading are handed to smaller, cheaper models, and each chat only loads the tools it needs. Your limit goes on the real work.
Git worktrees
Each chat can work on its own copy of the project and knows what the others are in the middle of, so two agents never rewrite the same file. When one is done, Finish saves its work for you.
Permission prompts
OrcStrator spots the safe things that keep asking and lets you allow them once. The risky ones still stop and ask, in plain words.
MCP servers, skills and plugins
Switch tools, skills and plugins on and off per agent with a click, so each chat loads only what the job needs.
Terminal commands
No copying commands into a black window. You see what it will do, press Run it for me, and anything risky is flagged in red first.
Cron jobs
Pick a day and a time and the job runs by itself, on your computer, with every run on a calendar.
Background processes
A chat only runs while it's working. The moment it stops, OrcStrator closes it, and your next message brings it back in about a second, cache and all.
I first built OrcStrator as an autonomous orchestration platform: agents running agents, no human in the loop. It was a massive failure, but I learned the best and the worst ways to save tokens. Slowly, over nine months, it became something better: an operating system for founders and professionals, with you at the wheel.
| Typical multi-agent tools | OrcStrator | |
|---|---|---|
| How Claude runs | Behind their own layers: MCP servers relaying messages between agents, bridges, sometimes a proxy in the middle | The official Claude Code terminal, started silently in the background. You never see it, but it's the real thing |
| Your keys | Many log into your Claude account, so you hand them your keys | Never touches them. Your keys stay on your PC |
| Extra tokens on every turn | Can be 20,000 to 50,000: tool lists, relayed messages, a dump of everything going on | About 1,400, sent once when a chat starts, then cached |
| A 20-message chat on Opus 5.5 | $1.60 to $4.00 of extra input, before Claude writes a word | About 2 cents |
| Chats you're not using | Stay open as running programs, eating your memory | Not running at all. 20 chats open, 2 working: 2 use your computer |
The extra-token cost is at Anthropic's API prices for Opus 5.5 ($4 per million input tokens, $0.20 when cached), for overhead that changes every turn and so can't be cached. On a Claude plan you don't pay per token: the same tokens come out of your limit instead.
Red flag
A lot of tools log into your Claude account, so you hand them your keys. OrcStrator never gets your keys. They stay on your PC.
Open Claude Code the normal way and every terminal stays alive while you're not using it, holding on to your computer's memory. OrcStrator plugs into Claude Code and listens to what it does. The moment a turn ends and Claude stops working, OrcStrator closes that process on the spot.
Those 18 quiet chats are still alive, just not on your computer. The conversation and its cache live on Anthropic's servers, so your next message starts a fresh process that resumes the same chat with the cache still warm, and it takes about a second.
Back in February I spent a couple of hours exploring Claude Code. It was so hard to use, so expensive, and so easy to make mistakes. There were so many commands I didn't want to learn. I wanted buttons I could click.
So I spent the next 57 hours streamlining everything I could, and built the first version of OrcStrator. Then I spent two weeks rebuilding my whole system with it. That system became a juggernaut, and agencies now pay me every month to use it.
I also built and improved OrcStrator using OrcStrator. And it's how I built the Company OS my business runs on.
I never planned to sell it. I was showing it to clients, and then clients started using it. About 10 of them ran an earlier, patched-together version before this one existed.
Then my wife started using it, with zero technical background, and her output was incredible. If it works for her, it works for anyone who would rather click a button than learn a terminal.
Anthropic's terms of service are 100% respected. Anthropic only allows a Claude subscription to be used through Claude Code and claude.ai, and OrcStrator runs the official Claude Code, on your own login. Tools that plug your subscription into something else can get your account shut down at any moment: people who connected their Claude plans to OpenClaw found that out the hard way. I don't know of any other tool like OrcStrator that stays inside the rules.
I took care of that on purpose: my entire business runs on Claude Code, and I can't carry the risk of losing my account because a tool broke a policy. You get the same protection. I won't lose my business, and you won't lose yours.
Everything Claude Code does with your data is covered by Anthropic's own privacy policy, because it's the official Claude Code doing it. One of my agents keeps checking that OrcStrator still meets all of it, and if anything changes, OrcStrator changes with it.
It's not meant to be nice. It's meant to be strong.
The community
Where OrcStrator users share how they actually run their business with it. People learning to build and improve a real business with AI, not random dudes trying to make their first dollar online.
Or you could just not.
The top plan turns the board into a project manager that works while you're in meetings, asleep, or pretending to be in a meeting.
Ask in plain words ("start the weekly report", "what's stuck?") and The Orc makes the cards, handles the routine questions and writes a timeline of what it did.
Anything you repeat becomes a routine: every Monday at 9, every day at 6. It fires by itself and shows every run on a calendar.
When your Claude plan runs out mid-job, chats carry on with Haiku or Sonnet on your own API key. The work doesn't stop because a counter said so.
Talk to The Orc on Telegram. When an agent needs a yes, your phone gets Accept and Reject buttons. Approve things from the supermarket queue.
Start free. Upgrade when your agents outnumber your excuses.
$300/month
Everything in OrcStrator, plus the operating system your company runs on. Your CRM, every client and prospect, your emails and the follow-ups that send themselves, all built in and run by the same agents that already work your board. Bring your team: everyone works the same board. One place for the whole business, not twelve tabs and a spreadsheet nobody updates.
Get early access* Built on Claude Code's own features (/mcp, /skills, /plugin). They come with your Claude plan, they work in the terminal, and they work in OrcStrator on every plan. What Pro adds is the management on top: switching them on and off per agent, so each one loads only what it needs.
OrcStrator runs on your computer, so your computer has to be on (with OrcStrator open) for agents and routines to work. They're good, not magic.
Plans cover OrcStrator only. Claude usage comes from your own Claude subscription, as it always did. Claude's own cloud routines stay yours too; OrcStrator routines are separate and run on your machine.
Click Download for Windows and open the file. The installer sets up everything it needs, Claude Code included.
OrcStrator opens with a Sign in page. Pick Create account, add your email and a password. Then sign in to Claude the way you normally do.
Click the link in the email I send you, sign in, and you're on the board. That's the last login page you'll see unless you sign out.
Windows shows a blue warning for apps that aren't code-signed yet. OrcStrator isn't, for now. Click More info, then Run anyway. This goes away once the app is signed. Until then, Windows is just being Windows.
That's who it's for. You write what you want in plain words and read the results in plain words. The code stays somewhere you never have to look.
Yes. OrcStrator runs Claude Code on your own Claude account (Pro or Max). I don't sell Claude usage and I don't sit in the middle of it.
No. Chats, files and projects stay on your computer. Your work goes straight to Claude on your own account. I only see your email, a scrambled id for your computer and your plan.
Windows only for now. Mac is on the list.
Not for this. OrcStrator runs the official Claude Code on your own login, which is exactly how Anthropic allows a subscription to be used. It never copies your login into anything else.
The desktop app is one chat at a time. OrcStrator runs many Claude Code agents side by side, on a board you can see, with routines, setups and limits on screen, and it works on the files on your computer.
The opposite. Chats stay cached, big ones get summarized, and searching goes to cheaper models, so the same plan goes further than typing into Claude Code yourself.
You see it coming on screen, and you see exactly when it resets. Nothing is lost: chats pick up where they stopped.
Yes. The agents work on your computer, so it needs to be on and awake for them to run, routines included.
It only works in the folders you give it, risky actions stop and ask you first in plain words, and agents working on the same project each get their own copy, so they don't overwrite each other.
Nowhere, if I can help it. When you close a chat, OrcStrator deletes anything it can identify as a password or key from the saved transcript on your computer.
Yes. It's the same Claude Code underneath, so everything you already set up keeps working. Setups let you switch them on and off per agent.
On the top plan, yes: talk to your agents through Telegram, get their questions and finished work there, and answer with a tap.
Yes: AI Marketing and The Orc, where people share how they actually use it to get and keep clients.
Soon. Want early access? Email nask@rodrigonask.com.
Hundreds of routine runs this week, five agents working, chats waiting on a yes, and every routine of the week below. Demo data, real app.
Free plan. No card. Runs on your computer.