Describe it. Generate it. Own it.
Nx·Studio is a web app for designing digital workers. Describe a business process and an AI analyst proposes up to five ways to automate it. You adjust the one you choose, test it on made-up data, and download it as software you run yourself: a complete project, source code included.
- No license check
- No usage meter
- No calls back to Nexgile
The Studio designs and tests. It never runs your live work: the worker you download runs on your own systems.
Your description
“Match invoices to purchase orders, route exceptions to finance, and escalate anything unusual.”
- 01 Describe the process
- 02 Get up to five designs
- 03 Shape it on a canvas
- 04 Test it on made-up data Testing
- 05 Download it
- 06 Run it yourself
An illustration of the journey through the Studio, not a screenshot.
What Nx·Studio is, and why teams use it.
A design studio for digital workers.
A digital worker is an automated workflow: AI steps, fixed rules, connections to your systems, and points where a person must approve. Nx·Studio is the web app where you design one, test it and download it.
Automation that takes months, then locks you in.
Automating a process usually means a long specification and a tool you rent for as long as you use it. Nx·Studio turns a description into a tested design, and what you take away is software you own.
The people who know the process, and the people who run it.
Process owners and analysts describe the work and review the design. Engineers refine it and run what is exported. Building and editing a design needs no code; running it in production is an engineering job.
From a description to software you own, in six moves.
- 01
Describe the process
Answer a guided questionnaire, paste your notes and documents, or write a single paragraph about how the work is done today.
- 02
Get up to five designs
An AI analyst proposes up to five complete designs for the worker, from cautious to ambitious, and scores each one. You choose.
- 03
Shape it on a canvas
The design is a diagram of steps. Drag, connect and adjust them without writing code, while the Studio checks your work.
- 04
Test it on made-up data
The Studio creates test records shaped like your process. Run the worker, watch each step, and approve or reject when it pauses for a person.
- 05
Download it
One button downloads the worker as a complete software project, with a security review of the design and a signed list of everything inside.
- 06
Run it yourself
Your engineers start it on your own servers or cloud. It carries no license check, no usage meter and no call back to Nexgile.
Tell it about your business, your way.
The more Nx·Studio knows about how you work, the better its designs. It asks about your team, your systems, your rules and your limits, and afterwards shows whether each rule made it into the design.
Three ways to begin
A guided ten-step questionnaire, a quick paste of your notes and documents, or a single paragraph.
Eight starter packs
Prewritten starting points, from month-end close to IT incident triage, that you edit to fit your company.
Start from your event logs
Upload a log of who did what and when. The Studio maps the real process and ranks where a worker would help most.
Your rules, tracked
State your hard rules and approval limits. After generation, each one is listed as honored or not.
Design it like a diagram.
The canvas shows the worker as connected steps that anyone can follow. Add, remove and reconnect them, and the Studio checks the design as you go.
Prefer a business view? The same worker can be shown as lanes by owner, person, team or system, so a reviewer can follow it without the technical canvas.
40 kinds of step
AI reasoning, fixed rules, approvals, connections to other systems, document handling and more.
16 workflow patterns
Common shapes, such as a pipeline with checkpoints or a review by a person, dropped onto the canvas in one move.
Checked as you build
Structural mistakes and missing settings show up while you edit. A design that breaks a hard rule cannot be saved.
Versions you can compare
Every save is a numbered version. Compare any two to see exactly what changed, and merge them.
Tested before it touches anything real.
Nx·Studio designs and tests. It never runs your live work. You try the worker on made-up data until you are satisfied, then export it.
Test data shaped like yours
The Studio generates fictional records that match the worker's own data, or you import a file of your choosing.
Watch every step
Run a scenario and see what each step did, how long it took and what it cost. Step back through a finished run to find a problem.
Approvals behave like the real thing
When a test run reaches an approval step it stops and waits for a person to approve, reject or edit.
Scores and a report
Run sets of scenarios and score the worker for speed, consistency and coverage, and for accuracy against the checks you define.
Each evaluation can be downloaded as a report written for procurement and security review. A sample is available on request.
Ask for a sample reportFixed rules for routine decisions. AI where judgment is needed.
A digital worker mixes the two. The Studio shows you the mix, and what it will cost, before you export anything.
Rule steps never call AI
A decision designed as a rule gives the same answer for the same input, every time, and adds nothing to your AI bill. Type a rule in plain English and the Studio writes the logic.
See the mix and the cost before you export
Each design shows what share of its steps are fixed rules and what share are AI, with an estimated cost per run and per month at the volume you expect.
Limits on every AI step
Set the response size, the number of retries and the time allowed for each step, and a time limit for the whole run. Spending budgets are recorded as guidance today; they are not yet enforced.
“The cheapest AI call is the one that never happens.”
Once exported, there is no meter. The AI your worker still uses is billed by your own provider, at your rates.
A design has to pass its checks before it can be exported.
A worker's design is something you can read: its steps, its rules, who approves what, and where work goes to a person. The Studio checks the structure of that design and can refuse to export one that breaks a safety rule.
You can also write your own rulebook, for where data may be kept, how long, who may access it and what must be logged, and see how far each design satisfies it.
- Invoice intake: match against the purchase order
- Routine routing: a fixed rule, same answer every run
- Refunds: one approval step, any amount
- Low scores: sent to a person
- Invoice intake: match against the purchase order
- Routine routing: a fixed rule, same answer every run
- Refunds above $10,000: a second approval step
- Low scores: sent to a person
Illustrative: two versions of a design compared in plain language. When a policy changes, you can see exactly what changed in the worker.
The four checks that can stop an export
A person before any outside action
A step that acts on another system must come after an approval step.
Sensitive data stays in
Personal or health data must pass a redaction step before it can leave.
No dead ends
Every decision point has to cover every case.
Records in order
Audit events must be written in the right sequence.
“Is it really ours?” Read the code and check the record.
The export is a complete software project in Python: the worker's design, the code that runs it, and the files to start it in a container. Everything arrives as readable source code.
Each download lists what is inside it and carries a signature over every file. Your engineers run one script, offline, to confirm that nothing was altered after export. If you stop working with us, nothing stops working.
In every download
Always included- The worker's design, in one file
- The code that runs it, as readable source
- A start script and a container file
- Sample scenarios and a smoke test
- A threat model written from the design
- A bill of materials: every model, tool and software part inside
- A signed record of every file, with a script that checks it offline
Yours to add
Optional at export- Starter set-up files for AWS, Azure, Google Cloud, Databricks, Snowflake or Kubernetes
- Four monitoring dashboards and a set of alert rules
- A generated operating guide and architecture documents
- Monitoring for drift from tested behavior
- Redaction of personal data
- Safety checks on what goes in and what comes out
- Long-running work that survives a restart
- A web chat page, or an address other AI tools can call
- The worker published as a tool for other assistants
Can Nexgile still get in?
No. The download has no Nexgile address, key or license check in it, and it runs where you put it.
Can I export it, or am I stranded?
Exporting is the product. You download a complete Python project with a container file and a start script.
What if Nexgile goes away?
It keeps running. Nothing in the download depends on us.
Connect it to what you already run.
A worker reaches your other systems through open, standard routes, with credentials that never leave your side.
Web services
Call any system that offers a web API, using credentials that stay in your own vault.
Tool servers
Use any tool published through the open standard for connecting AI to other systems.
Databases
Ask a question in plain language and the worker turns it into a single read-only query.
Documents and memory
Give a worker documents to consult, a map of how things relate, and memories to start from.
Older systems without an API
The worker writes a structured request for the automation (RPA) tool you already run, which carries it out.
Credentials stay with you
The Studio stores only the name of a secret, never its value.
What is ready, what is in preview, what is still being built.
We would rather you know now. Everything on this page is in the product today, unless it appears in the second or third list below.
Everything described above this section, including:
- Describing a process in three ways, with eight starter packs
- Up to five generated designs with scorecards
- The canvas, its checks, versions, compare and merge
- Test data, test runs, approvals that pause, evaluation reports
- Export checks, the download, its threat model and signed record
Built, and switched off by default:
- Changing a design by chatting with an assistant on the canvas
- Describing a process out loud
- Running a set of hostile inputs against a worker
- What-if runs side by side
- A guided presentation of a test run
- Rebuilding a design from existing automation code or a process diagram
- Updating an exported worker while keeping your engineers' changes
Partly built. Do not plan around these yet:
- A reviewer inbox with notifications by chat and email
- Two-person approval and escalation when a deadline nears
- A tamper-evident audit log as a standard part of every export
- Working calls to the catalog of about 3,300 public API definitions
- Several people editing one canvas at the same time
- One-click return to an earlier version
Bring one process. Leave with a design you can test.
See a digital worker generated from a description, tested on made-up data and exported, live.
You generate it. You preview it. You own it. You run it.
Nx·Studio designs and generates digital workers from a description of your business. Nx IDE is the code editor your engineers use to build, maintain and extend software. What the Studio delivers is ordinary source code, so it opens in Nx IDE like any other project.