What it does for your process. What your team gets back.
Start from a process you run, such as invoice matching or claims intake. See what it would cost to run, the hours it could save, how it is tested first, who approves what, and what you keep.
Weigh cost and time saved per design.
Describe a process your team runs today, such as supplier onboarding, or adapt a starter pack. Nx·Studio’s AI analyst returns as many as five options. Each shows the steps, connected systems, review points, estimated operating cost and hours saved, as planning inputs to test against your volumes.
Rules you can rely on
Your written policies, such as approval limits or payment terms, become rule-based steps wherever they can be checked. Those steps give the same answer for the same case, with no AI call and no AI cost. You accept or edit each drafted rule, and the Studio lists the rules that could not become checks.
About the FoundationFitted to your operation
Your industry, team, systems, policies and targets shape each design, so it fits the way your operation already runs.
Pick your starting point
Use whichever brief suits you: the 10-step guided form, a pasted website with notes, a single paragraph, or any of the eight starter packs. Documents such as procedure manuals can be attached through the form or the paste option. Optional modules, inactive until enabled, can study process logs and existing automation code. Shared material serves design work only, and a data-processing agreement with us is signed before any of it is shared.
Up to five options to weigh
Every option is scored and shows who does what, the steps, the systems it touches, and its estimated operating cost and time saved. Options that fail the Studio’s checks are corrected or removed from the list you receive.
Changes in your own words
Optional: type a change request such as “route refunds above a limit to a supervisor”. The assistant proposes the edit, and you see exactly what changes before accepting or rejecting it.
A shortlist ranked by estimated operating cost and time saved, which you sense-check with your own volume and support figures before committing budget.
Tested before it touches live work.
Nx·Studio can generate realistic sample cases, such as invoices or claims, for the process a worker handles, which lets you watch its behavior before deployment. Step through a run or execute it end to end, with genuine approval pauses. A sped-up simulation can exercise a group of workers together. Adversarial testing is optional and returns a scorecard. Every version lists its changes, and no real customer or supplier record is needed.
Rehearsal before reality
Previews, simulations and optional attack tests (red-teaming, where testers try to break the worker) all run without production data. Agree confidentiality and data-handling terms with us before you share process information.
About the FoundationMatching sample data
Generated records mirror the inputs a worker uses. Your own samples can be imported as well, up to 200,000 rows per dataset.
Scores you can compare
Each test run reports scores for accuracy, speed, coverage and robustness (how it copes with unusual cases), so you compare options on evidence.
Two runs, compared
Line up a pair of runs, or open an earlier run’s record and step through it.
Optional attack testing
Red-team tests are opt-in and end in a scorecard. Separately, the simulation lab flags conflicting updates and broken hand-offs between workers before rollout.
Test results give your security and operations colleagues evidence to review. You decide when the worker is ready for real cases.
Know who approves what, in advance.
A versioned blueprint, the worker’s documented design, defines every worker. Before production use, your governance colleagues compare, test and review its versions in a visual diagram or in swimlanes that name each step’s owner. Your auditors examine this blueprint as the record, and you receive it as your own code, not as a dashboard you rent.
Blueprints you can read
The blueprint captures what the worker does, which steps need a reviewer and how cases move, in one versioned design instead of loose AI instructions. IT manages its versions like any software, managers read the swimlane view, and your auditors get a plain-language report of the build.
About the FoundationSaved versions
Nothing is overwritten: every change you save is kept, ready to compare with or merge into another. Teams can return to a checkpoint they named earlier, and every test score shows which version earned it.
Review points by role
Each review point names who decides, such as a team lead or a finance approver. Your IT team connects the reviewer inbox app that ships with the export; Slack delivery can be added when you export.
Audit evidence
Every approval decision is recorded. Optional extras: a tamper-evident run log, and lineage that ties each field to its source.
Review the audit-trail specificationMinimum score for AI steps
Set the minimum score an AI judgment must reach. Results below it are flagged by default, or routed to a reviewer if you configure that. Sensitive actions without a review point are flagged as well, and every review point keeps the worker waiting for a decision.
Auditors can assess a worker’s behavior before it runs, using a record that belongs to you.
What you own, and what IT runs.
Nx·Studio exports the complete software, ready for your IT team to deploy, not a slide deck. Each export contains starter code for connecting to your systems, a reviewer inbox app, tests, and optional safety checks and monitoring. It runs wherever IT chooses to host it. There is no vendor runtime, no fee per decision and no call home to Nxgyle.
Request the code-quality bundle from the Trust CenterProof in every export
Every export includes a signed inventory of everything inside the software, which your IT team can verify offline. It also contains a security analysis for your IT security team (a threat model) and an EU AI Act template for your compliance colleagues to complete. The Studio also produces an ownership report; ask the Trust Center for a sanitized sample.
About the FoundationA catalog of integration connectors and starter code
SAP, Oracle EBS, Workday, Salesforce and ServiceNow connections come as starter code that your IT team finishes with each system’s access details. Links through APIs or AI gateways also need compatible interfaces, credentials and testing.
Inventory IT can verify
The signed inventory lists the software and AI components in the package, and your IT team can verify it without a network connection.
Review points that pause
At each review point the worker stops until a person approves or rejects. Reviewers decide in the export’s inbox app once IT connects it, or in Slack if it was enabled at export.
Risk safeguards
A security analysis (threat model) ships with each export. Personal-data redaction, policy checks that block disallowed actions, and monitoring for drift (results that shift over time) are optional.
Hosting choice
IT can run the software on AWS, Azure, Google Cloud, Databricks, Snowflake, container hosting or your own servers. It uses the AI model service you select, requires no license key, and does not depend on Nxgyle at run time.
Exported workers run without any Nxgyle runtime, so plan hosting, support and maintenance as internal costs.
Weigh three sourcing options.
Each option carries real trade-offs. Here is how we assess all three, including our own.
Ask IT to build on open frameworks.
You keep the code, while your IT team carries design, integration, controls and long-term upkeep.
License a hosted agent service.
Compare hosting, visibility, pricing and exit terms in the service agreement before you sign.
Generate it. Keep the code.
Designed around your process, measured on sample data, and handed to your IT team as source code you hold. This is the option we built.
What you can show reviewers.
Every capability here leaves paperwork your risk, audit and IT colleagues can check. Two illustrated samples follow.
- Intake purchase requests from the procurement portal
- Check budget code, supplier status, policy limit
- Approval requests over budget go to one manager
- Approval requests over budget need a manager and finance
- Route incomplete requests return to the requester
Illustrative. A policy change, such as a new approval limit, checked before rollout and traceable afterward.
Blueprint details on the platform page-
Scenarios run on sample data
Step by step or in one pass, with no real records involved.
-
Optional attack tests (red-team)
Adversarial scenarios, if you enable them, run before rollout.
-
Review points held for a person
Each one holds the worker until someone approves or rejects.
-
Runs compared, step by step
Two runs side by side, showing what changed between them.
Produced from a sample-data run and formatted for your procurement file (RFP).
Ask for the example report via Nx·StudioPublished catalog coverage.
Counts reflect published catalog records, not customer deployments or measured business results. Each digital worker runs a defined business workflow under configured rules and approvals. Nx·Studio exports its source code, which your organization then operates and maintains.
One capability, two long-term outcomes.
Compare what each model leaves you with when the contract ends.
Set by the provider’s agreement.
Subscription and usage terms apply.
Only what the provider makes visible.
Its terms decide what remains.
Servers you choose.
Rules handled without AI costs.
Blueprints reviewed before rollout.
You keep the code and its upkeep.
We charge for design and delivery; running the workers carries no Nxgyle fee.
Our model favors your independence, and ongoing support stays optional.