Every cluster its own window
Acceptance on the left, production on the right. The coloured edge tells you which cluster you are in, and production is red.
Studio now runs on your computer. With an assistant that builds alongside you.
The environment you build in on Core is now an app. You say what a flow should do and the assistant puts a draft on the canvas. Nothing goes live until you publish.
Watch the assistant
demo / API ordersEdited 2 minutes ago
AI working in this blueprint
demo - AIapiOn API GET
orders/open
v1
No rate limit
No throttle (lock & wait)
scopes (1)
copy url
On Request
demo - AIfunctionexecute
f(x)open orders
demo - AIfunctionprivate
open orders
pin
Connection
Params
Query
demo - AIcustomvalue
open
demo - AIquery: termfields
boost
name open
demo - AIconnectioncurrent request
demo - AIstoragesearch
Query
Limit
Offset
Cache Key
Storage
Results
hits
guids
Has hits
Profile
success
error
cursor
demo - AIcustomnumber
200
demo - AIresponsereply in JSON
array
code 200
type
Headers
fields
array
options
New conversationBuild an endpoint that lis…
What are we building?
Ask about this cluster, or describe what you need. Nothing goes live until you publish.
- Show me what is on this cluster
- Build a new flow
- Tell me what changed recently
Build an endpoint that lists the open orders
Thinking
I will look at how orders are stored first.
Looking through the blueprints
Read 3 things on the cluster, looked up 1 topic
Orders live in the orders storage. I can build GET orders/open in this blueprint, with a query that keeps only the open ones.
Working, 4 sWorking, 9 sWorking, 17 sWorking, 26 sWorking, 38 sDone in 38 s4 steps
- Opening the flow as source
- Editing the source
- Staging the change
- Running a test
API ordersBlueprint
Draft, not live yetDiscardPublish
PublishedUndo
Publishing
Published. GET orders/open now answers on demo.
Start this plan?demo
In the assistant's wordsBuild an endpoint that lists open orders.
- Changes: API orders
- Tests stay inside the cluster.
- Nothing goes live until you publish.
- Publish when it is ready, without asking again
- Don't ask next time on demo
Not nowStart
Publish this draft?demo
- Blueprint: API orders
- Draft: 9 new blocks.
- This goes live as soon as you publish.
CancelPublish
API orders
Build an endpoint that lists the open ordersAsk, or say what to change
Shift + Enter for a new lineClaude
demo / API ordersEdited 2 minutes ago
AI working in this blueprint
apiOn API GET
orders/open
v1
No rate limit
No throttle (lock & wait)
scopes (1)
copy url
On Request
functionexecute
f(x)open orders
functionprivate
open orders
pin
Connection
Params
Query
customvalue
open
query: termfields
boost
name open
connectioncurrent request
storagesearch
Query
Limit
Offset
Cache Key
Storage
Results
hits
guids
Has hits
Profile
success
error
cursor
customnumber
200
responsereply in JSON
array
code 200
type
Headers
fields
array
options
New conversationExplain this flow
What should this flow do?
Ask about the flow you have open, or say what it should do differently.
- Explain this flow
- Change this flow
- Find out why a request failed
Explain this flow
Reading the flow
Following the wiring
Read 2 things on the cluster
This flow is one endpoint, GET orders/open. It lists the orders that are still open.
On API GET registers the endpoint and checks the caller's scope, then hands the request to the function open orders.
search reads the orders storage. The query on the left keeps only orders whose state is open.
reply in JSON answers the request with code 200 and the results.
API orders
Ask, or say what to change
Shift + Enter for a new lineClaude
demo / API ordersEdited 2 minutes ago
apiOn API GET
orders/open
v1
No rate limit
No throttle (lock & wait)
scopes (1)
copy url
On Request
functionexecute
f(x)open orders
functionprivate
open orders
pin
Connection
Params
Query
customvalue
open
query: termfields
boost
name open
connectioncurrent request
storagesearch
Query
Limit
Offset
Cache Key
Storage
Results
hits
guids
Has hits
Profile
success
error
cursor
customnumber
200
responsereply in JSON
array
code 200
type
Headers
fields
array
options
New conversationFind out why a request fai…
What should this flow do?
Ask about the flow you have open, or say what it should do differently.
- Explain this flow
- Change this flow
- Find out why a request failed
Find out why a request failed
Looking through the blueprints
Reading the flow
Read 4 things on the cluster
One request to GET orders/open was refused at 14:02, with a 403.
It never reached the flow. On API GET asks for the scope orders.read, and the token that called it only had orders.write.
Nothing is wrong with the flow itself. Give that token the orders.read scope and the same request goes through.
API orders
Ask, or say what to change
Shift + Enter for a new lineClaude
The assistant in Studio, drawn from the real app, on a demo cluster.
Say what the flow should do. The assistant builds a draft.
The assistant sits in a panel beside your canvas. It reads along in your cluster, proposes a plan and puts the result in your blueprint as a draft. What it makes are ordinary blocks, yours to check and change.
-
API orders3 selected blocksBuild an endpoint that lists the open ordersShift + Enter for a new lineClaude
You say what you need
In plain language. The assistant knows which flow you have open and which blocks you selected.
-
Start this plan?demo
In the assistant's wordsBuild an endpoint that lists open orders.
- Changes: API orders
- Tests stay inside the cluster.
- Nothing goes live until you publish.
Not nowStartYou start the plan
You see up front what will change and whether anything gets tested. Nothing happens until you press Start.
-
demo - AIapiOn API GETorders/openv1No rate limitNo throttle (lock & wait)scopes (1)copy urlOn Requestdemo - AIfunctionexecutef(x)open orders
The draft appears on your canvas
Outlined with dashes and signed with a name. You and your colleagues see what is new and where it came from.
-
API ordersBlueprintDraft, not live yetDiscardPublishPublishedUndo
Published.
GET orders/opennow answers on demo.You publish
Or you discard the draft. Until then nothing is live.
On the plan you already have
The assistant runs on Claude Code or Codex, on your own computer. You sign in with your own Claude or ChatGPT account and choose the model yourself. Usage counts against the plan you already have.
Claude: Pro, Max, Team or Enterprise. ChatGPT: a plan that includes Codex. On macOS for now.
Claude
- Claude is installed on this computer
- Signed in with your Claude account
- Checked with this version of RUAL Studio
ChatGPT
- Codex is installed on this computer
- Signed in with your ChatGPT account
- Checked with this version of RUAL Studio
Nothing goes live until you publish.
The assistant may read and prepare a great deal. You decide, in a place it cannot reach.
- The question comes from Studio itself
- You confirm starting and publishing on a card the app draws, outside the conversation. Nothing the assistant reads or writes can answer that card.
- A test that can reach outside asks first
- Tests that stay inside your cluster run on. If a test could send mail or call another system, you decide.
- Every step can be read back
- What was asked, what came back and what changed in the source.
- Your account, your data
- What you ask, and what the assistant reads from your cluster to answer it, goes to Anthropic or OpenAI under your own account.
Publish to production?production · live
- Blueprint: API orders
- Draft: 9 new blocks.
- This goes live as soon as you publish.
CancelPublish
Publish this draft?demo
- Blueprint: API orders
- Draft: 9 new blocks.
- This goes live as soon as you publish.
CancelPublish
⌘ Enter to answer
A window per cluster, and Studio keeps you posted.
As an app, Studio does what a browser tab cannot: several clusters side by side, your sessions stored safely, and a nudge when something happens while you are busy elsewhere.
Status in your menu bar
A dot per open cluster: green when it is reachable, red when it is not.
API ordersChanged by Mara while open in your window.
2mDeploy finishedAPI orders deployed successfully.
nowA notification when your deploy is done
Also when a colleague changes the blueprint you have open, or a cluster becomes unreachable.
Recent Sessions
- Acceptanceacc.yourcompany.com
- Productionapp.yourcompany.com
- Testtest.yourcompany.com
You stay signed in
Your sessions are stored encrypted on your computer, using your system's key store.
storagesearch
Query
Limit
Offset
Cache Key
Storage
Results
hits
guids
Has hits
Profile
success
error
cursor
responsereply in JSON
array
code
type
Headers
fields
array
options
Have a look at this blockrual://acc.yourcompany.com/blueprints/view/8f3a…?a=search
A link opens the right block
Copy a link to a block and your colleague lands on it with one click.
Keep working, even without a connection.
Studio keeps a copy of your blueprints on your own computer. If the connection drops, you carry on building. As soon as you are back online, Studio updates the cluster.
- Blueprints open instantly, from your own disk
- Changes are kept until there is a connection again
- If your change collides with a colleague's, you choose
functionprivate
open orders
pin
Connection
Params
Query
storagesearch
Query
Limit
Offset
Cache Key
Storage
Results
hits
guids
Has hits
Profile
success
error
cursor
customvalue
open
query: termfields
boost
name open
- 09:12The connection dropsStudio switches to the copy on your computer.
- 09:12 to 09:47You simply carry onYour changes are safe on your own disk.
- 09:47You are back onlineStudio updates the cluster.
Let your own agent work with you.
Already working with Claude Code or another agent on your computer, including one with an open model? Give it access to a cluster over MCP, through the session you are signed in with. It passes the same checks as Studio itself.
- Off by default
- You switch access on yourself and then share per cluster. Your system asks for confirmation both times.
- From your own computer only
- Nothing is reachable from outside.
- Everything in a log
- Every action an agent takes is recorded: which cluster, which tool and whether it worked.
- A workspace per cluster
- With RUAL's skills in it. Open it in Claude Code and your session knows how things work here.
› what does the API orders blueprint do?
• rual-demo bp_pull
• rual-demo studio_screenshot_blueprint
One endpoint: GET orders/open.
It lists the orders whose state is open.
Activity log
- demobp_pullok
- demostudio_screenshot_blueprintok
Still the workshop where you build all of it.
Everything Studio could already do is in the app too. And in the browser it all keeps working the way you are used to.
functionprivate
open orders
pin
Connection
Params
Query
storagesearch
Query
Limit
Offset
Cache Key
Storage
Results
hits
guids
Has hits
Profile
success
error
cursor
Build together, live
You watch your colleague's cursor move and blocks appear while they build. When an agent is working along, you see that too. Stuck? A RUAL engineer can look along remotely.
storagesearch
Query
Limit
Offset
Cache Key
Storage
Results[ {…}, {…}, 10 more ]
hits12
guids
Has hitstrue
Profile
success
error
cursor
Debug on the real thing
Click a block and see the data running through it. No log lines to add, no guessing.
search for a block…⌘K
- date104
- array99
- condition68
- files56
- query53
- users49
- http connection40
- anthropic27
- dobot22
- zpl21
- mollie18
- floriday8
2,800 blocks are ready and waiting
For logic, data and dates, and just as much for Exact, Mollie, DOBOT robots and ZPL printers. Missing a block? File a request and it is in within a week.
Tidy clears up your canvas
One key lays your blocks out neatly. You see what would move first, then decide whether to keep it.
CanvasCodeCompiles cleanly.
blueprint "Orders API" route POST "orders" -> create-order fn create-order(connection, params, query, body) { let {customer_guid, amount} = body guard amount > 0 else { reply 400 {error: "INVALID_AMOUNT"} }
Your flow as code
View a blueprint as Blueprint Language, edit the source and put the change back on the canvas.
TrafficLast 48 hours · 1,284 requests
- GETorders/open20014:06Replay
- GETorders/open40314:02Replay
- POSTorders20013:58Replay
Traffic and replay
See which requests your flow received and replay one. The original side effects are not repeated.
Full history
Every change to a blueprint is recorded: who, what, when. Looking back and rolling back come with it.
Comments on blocks
Discussion happens where it belongs: on the block itself, inside the flow, not in a separate mail thread.
Validation before go-live
Studio validates your blueprint and points out broken connections and missing fields before they reach production.
Deploy management built in
Want to roll changes out in a controlled way instead of straight to live? Studio has deploy management built in: stage it, review it, release it when you say so.
Time tracking built in
Development time is tracked automatically per developer and per project. You see where the hours go, without a separate timesheet tool beside it.
Custom blocks
Using a tool that does not have a block yet? We build that block into the platform and maintain it for you. Your team only has to drag it into place.
Block templates
For the patterns you use more than once.
Groups and copy/paste
For large blueprints.
Changes take minutes, not sprints.
Studio is approachable enough that the kid next door who was playing Minecraft yesterday is helping build your fleet management today. Because your flows are visual, everyone can see how the process runs, including your operations manager. You discuss a change at the screen and make it there and then. Internal questions resolve themselves before they turn into tickets.
- Change existing flows without a development project
- Insight into what your development team is building and where the time goes
- Usage statistics per environment, so you keep a grip on your IT
Who builds Studio
We are the heaviest user of Studio ourselves, and it shows in the pace.
By developers, for developers
We have been building business applications for fifteen years and we work in Studio every day ourselves. Anything that annoys us gets fixed, and every environment benefits.
Improvements within a week
Changes to the platform or to Studio are usually live within a week of approval. The changelog on docs.rual.nl updates live from our code deployments: what we build, you see appear there.
Download Studio.
For macOS, Windows and Linux. Updates arrive on their own, and Studio in the browser simply keeps working.
macOS
Apple silicon and Intel
- Version 3.1.2, 157 MB
- macOS 12 or later
Windows
Installer or portable
- Version 3.1.2, 130 MB
- Windows 10 or later, 64-bit
The assistant works on macOS for now. Already signed in in the browser? Open in RUAL Studio brings your sessions across.
Frequently asked questions
What do I need for the assistant?
A Mac, and Claude Code or Codex on your computer, signed in with your own account. For Claude that is a Pro, Max, Team or Enterprise plan. For Codex, a ChatGPT plan that includes Codex. Studio checks at start-up that everything is in place.
Where does my data go?
What you ask the assistant, and what it reads from your cluster to answer, goes to Anthropic or OpenAI under your own account. Studio never sees your password or token for that service.
Can the assistant put something live by itself?
Only if you tick that when you start a plan. It then publishes that one plan without asking again. By default Studio asks every time.
Will Studio in the browser keep working?
Yes. The app is an addition, the browser stays.
Does everything work on every cluster?
Tidy, the code view and replay need a recent version of Core. If your cluster runs an older version, Studio says so.
No cluster on Core yet?
In a demo you build a flow yourself. Then you know whether it fits your team.