Nuclearn CAP AI Book a demo
NEW The Smart CR workspace

Condition reports,
screened your way.

Nuclearn CAP AI helps anyone at the plant write a complete condition report in plain words, screens it against your own codes and procedures with the reasons shown, and keeps the actions, history and trends together. Your screeners keep the decision.

Book a demo See what it does
  • Uses your plant's form, codes and procedures
  • At your site or hosted by Nuclearn
  • Signs in through Nuclearn
Smart CR drafting a synthetic condition report: the assessment on the left passes when, while and what and flags who and the consequence, the plant's form in the middle holds the description, and five similar past reports are listed on the right.

Screenshots on this page show synthetic demo condition reports, as they appear in Nuclearn CAP AI today.

  • Backed by NuclearnAI software built for nuclear plants since 2021
  • Over 70 facilitiesNuclearn products serve sites in North America and the U.K.
  • ISO/IEC 27001:2022Certified information security management
Why teams use it

From the first report to the trend it belongs to.

A condition report starts with someone writing it down and ends with a screener deciding what it means. CAP AI helps with the work in between, using your plant's own form, codes and history.

CAP screening

Screen new condition reports

Significance, priority, operability and reportability, owner and trend codes, suggested for each new report with the reasons and sources behind them. Routine reports move on; the rest wait for a screener.

See screening → Anyone at the plant

Write a complete report the first time

Describe the condition in plain words and the form fills itself in, checked as you write against what your plant expects a good report to say.

See Smart CR → Initiators and screeners

Catch duplicates before they're filed

Similar past reports appear while you write and on the record, so a repeat condition gets linked to what's already known instead of being logged twice.

See similar reports → CAP and performance improvement

See trends while they form

Trend codes on each report feed a live dashboard of performance objectives and criteria, year over year, sliced by any field you need.

See trending → Owners and action holders

Keep the record moving

Actions, assignments, comments with @mentions, attachments and a revision history on one record, plus a feed of what moved on the equipment and units you follow.

See the record → Everyone

Raise a report from AtomAssist

Ask AtomAssist to write it up. It drafts the report into your Draft Box with the same checks as the form, and you review it and submit.

See AtomAssist →
The problem

Thousands of reports a year, each one read by hand.

A healthy corrective action program depends on a low threshold for reporting, so the reports keep coming: equipment, procedures, people, paperwork. Each one has to be read, classified, coded and assigned, and a screening meeting settles the hard ones.

When a report leaves out what was going on, who found it or what it affects, the screener goes back to the person who wrote it. When nobody spots that it was reported last month, it's logged twice.

CAP AI takes the routine reading off your screeners' desks and helps the person reporting get it right the first time.

6,0001
condition reports per reactor unit, per year

INL's conservative estimate, from a survey of condition report datasets from multiple utilities.

$1.5M1
per reactor unit, per year, to screen them

INL's estimate of what screening costs today: checking whether a condition is already reported; judging criticality, operability, significance and severity; each department's validation; and the screening meetings. Online work only, at $200/hr.

INL's figures are its estimates of what these steps cost now, not savings from Nuclearn CAP AI; its step times are expert estimates and vary by plant. Outage work is not included. Sources at the foot of the page.

How it works

Fits the way your CAP already works

Start with screening on the system you have, or bring intake and the record into Nuclearn with Smart CR.

Report the condition

Write it in Smart CR, raise it from AtomAssist, or keep using your current CAP system. CAP AI picks up new reports as they arrive.

CAP AI screens it

Your plant's screening template suggests each field from the narrative, your procedures and similar past reports, and shows why.

Your screeners decide

Reports inside your thresholds move on. The rest land in a screener's queue, and a sample you set goes to a person regardless.

Screening

Screening you can check, report by report.

For each new report, CAP AI suggests the screening fields your plant uses: significance, priority, operability, reportability, owner, trend codes and more. Fields inside your thresholds move on; the rest go to a screener. The audit history shows which reports were screened automatically and which fields went to a person, with what the model was given and how confident it was.

Moves onFields inside your plant's confidence thresholds.
To a screenerAnything below threshold, and any report picked for the quality sample you set.
GroundedWith procedure-grounded screening, each value shows why and cites your procedures, approved conventions and similar past reports, with quotes checked against the text.
OverriddenA screener picks another label, and the record shows who decided.

Screening decisions stay with your plant: you set the thresholds and the sample, and only people with screening authority can record or change a decision.

CAP AI's audit history for a screening template: synthetic condition reports with a tick where a report was screened automatically, whether the result matched the screener's decision, and the predicted category, with a person icon where the field went to a screener.
Audit history: what was screened automatically, and which fields went to a person.
Smart CR intake

A complete report on the first try.

Describe what you found in your own words. Smart CR drafts the rest of your plant's form as you type and grades the description live against what your plant says a good report contains: by default, when, while, who, what and the consequence. Before you submit, it lists what's still to confirm.

  • Similar past reports appear as you write, before anything is filed.
  • Find the equipment in your plant's equipment list, and tie the report to a controlled document.
  • Attach a photo, and Smart CR can draft the report from what it shows.
  • Ask the assistant beside the form what a screener will look for.

Suggested classifications are marked as suggestions: you decide, and the screening committee finalizes. On submit, the report goes to your plant's dataset and is screened by the template your administrator chose.

The lower half of the condition report form: the equipment tag with a Look up button, the affected system code, immediate actions taken, a controlled document field marked needs input with its own Look up, and Attach photo. The plant's criteria settings: what a good report says about when, while, who, what and the consequence, which the assistant grades each description against.
Lookups on the form, and the criteria your plant grades descriptions against.
The record

One record, from the report to its actions.

Each condition report has one record in the Smart CR workspace: the report, its screening decisions and evidence, the actions it owes, the discussion and the files. Assign it to an owner group and pull a colleague in with an @mention.

  • Actions for evaluations, action items, work orders and commitments, with due dates, work-order references, apparent cause, extent of condition and cause codes.
  • A history of who did what, in order, and a PDF/A snapshot of each revision.
  • Fields open and lock by step, so what was reported stays as reported.

Coming next: the review, closure and effectiveness-review steps of the record's lifecycle, set up for your plant the same way.

A synthetic condition report record: its progress from Initiate through Screening, tabs for report details, screening, actions, discussion and history, the owner-group assignment, and the report as written. The record's history: the audit trail of who did what, in order, and a PDF/A snapshot for revision 1 with a download link.
The report, its screening, its actions and its history in one place.
The workspace

Follow what matters to you.

The Smart CR workspace opens on a feed of condition reports touching what you follow: a component, an owner group, a unit or a trend code. Reports assigned to you and ones where a colleague @mentioned you show up too, each with the reason it's there. Raise a new report from the same page.

  • Follow or mute anything, with suggestions from your plant's recent reports.
  • CAP reports by state and classification, with open and overdue actions and CSV export.
The CR workspace: a box to describe a new condition, an activity feed of synthetic condition reports with the reasons each is shown, and the components, owner group and unit being followed. The CAP reports panel: reports raised in the last 30 days, open and overdue actions, counts by state and by classification, and Download CSV.
A feed of what moved on the things you follow, and the CAP reports behind it.
AtomAssist and integrations

Raise a report from a conversation. Screen one from any system.

AtomAssist can write a condition report for you on your plant's form, with the same quality checks as the intake page. It saves a draft to your Draft Box and doesn't submit it: you review and submit. Your other systems call screening through the REST API.

AtomAssist + Smart CR
On rounds I found oil weeping from the outboard bearing housing on the B charging pump. Sight glass level looked normal. I told the control room. Can you write up a CR?
Draft_Condition_Report
I've saved a draft to your Draft Box: "Oil weeping from B charging pump outboard bearing housing." When, while, who and what read clean. The consequence is thin: say whether the pump stayed in service and whether the oil level was checked against its limit. I haven't submitted it. Open the draft to check it and submit.

Example conversation with synthetic details. The assistant drafts; a person reviews and submits, and screening starts only then.

Check my descriptionFor when, while, who, what and consequence
Suggest a clearer titleFrom what the report says
What will a screener look for?Before you submit
Which codes might fit?Classification and trend codes, as suggestions
Draft from a photoAttach an image to the report
Draft from AtomAssistSaved to your Draft Box, never submitted for you
# Screen one report with your plant's template
curl -X POST -H "X-API-KEY: $NUCLEARN_API_KEY" \
  -H "content-type: application/json" \
  -d @condition-report.json \
  "https://nuclearn.your-site.example/api/products/\
cap-screening-automation/automate?record_uid=$CR\
&automation_config_template_id=$TEMPLATE&automate=false"

# automate=false returns the suggested values to test with;
# automate=true records the screening for the template.

Access uses a Nuclearn platform API key, so the platform's sign-in, roles and licence apply.

Configured, not coded

Your form, your workflow, your trends

Automation KPIs

How many reports moved on and how many went to a person, how screening compares with your screeners' own decisions, and the manual sample, field by field and over time.

Your plant's form and criteria

Administrators set the fields a report carries and what a good description is graded against. The assistant is given exactly those fields, so the form stays yours.

Who may move a report

Each workflow step names who may take it. Left empty, it needs the record's own screening authority.

Action due dates

Default days for an action to close, set by classification.

Versioned definitions

Saved workflow definitions are kept, so an earlier one can be put back in force.

Promote when ready

Test a screening configuration, promote it to production and archive the old one, without a software release. Export and import configurations between environments.

Your platform's people and roles

Sign-in goes through the Nuclearn platform, so the same accounts and SSO apply, and platform roles and dataset permissions decide who can raise, screen, configure and read reports.

Deployment and security

Your reports stay in your Nuclearn deployment.

Condition reports, procedures and screening results live in your plant's datasets on the Nuclearn platform, and access goes through the platform's sign-in and roles.

At your site or hosted

CAP AI runs on the Nuclearn platform, on your own servers or in a deployment Nuclearn hosts for you.

Built for air-gapped sites

Outbound web access is an administrator setting, left off at air-gapped sites, and CAP AI's features are switched on in the deployment itself.

Nuclearn platform sign-in

Single sign-on through your identity provider, or the platform's own accounts. CAP AI keeps no separate user store.

Screening authority

Only people your plant gives screening authority can record or change a screening decision or move a report on.

Keyed access for systems

Your CAP system and in-house tools call screening and the record API with a Nuclearn platform API key.

A certified company

Nuclearn is ISO/IEC 27001:2022 certified. CAP AI, the Smart CR workspace and procedure-grounded screening are each switched on per deployment.

Ways to use it

Start with screening, or bring the whole report in.

Each part is switched on per deployment, so you can begin where your screeners feel it most.

Your current CAP system

CAP AI screening

Reads new reports from the CAP system you have, screens them with your template and writes the results back. Trend coding, the trending dashboard and KPIs come with it.

Screeners and analysts

Procedure-grounded screening

Screening that cites your procedures, approved conventions and similar past reports, checks its quotes, and escalates when the sources disagree.

Anyone who raises a report

Smart CR workspace

Plain-language intake, similar reports as you write, a record with its actions and history, and a feed of what you follow.

AtomAssist users

AtomAssist

Draft a condition report from a conversation, saved to your Draft Box for you to check and submit.

What it is not

CAP AI doesn't own your plant's corrective action decisions, and it isn't your QA program. It suggests, explains and keeps the record; your screeners and your program decide.

FAQ

Questions plants ask first

Do we have to replace our CAP system?

No. CAP AI screening works alongside the system you have: it reads new reports, screens them and writes the results back, and your other systems can call it through the REST API. The Smart CR workspace adds intake and a condition report record in Nuclearn if you want them; we'll talk through how that fits with what you run today.

Who makes the screening decision?

Your plant does. Under Appendix B to 10 CFR Part 50, corrective action is the licensee's program,2 and CAP AI keeps its decisions with you. You set the confidence thresholds and the share of reports sampled for a person to check. Anything below threshold, sampled or escalated goes to a screener, only people with screening authority can record or change a decision, and the record shows who decided.

Will it use our own codes and procedures?

Yes. Screening uses your plant's fields and labels. With procedure-grounded screening it cites your procedure datasets, your approved screening conventions and similar past reports. Past reports are context, not evidence on their own, and the report being screened is left out of its own precedents.

What does Smart CR count as a good report?

Whatever your plant says. Administrators set the form's fields and the criteria a description is graded against; the default is when, while, who, what and the consequence. The grade shows beside the form as someone writes, along with what's still to confirm before submitting.

What about cause evaluations, closure and effectiveness reviews?

Actions on a record already carry apparent cause, extent of condition and cause codes. The review, closure and effectiveness-review steps of the lifecycle are coming next, configured for your plant the same way as the rest.

Where does it run, and who can sign in?

On the Nuclearn platform, at your site or hosted by Nuclearn, including at air-gapped sites. People sign in through the platform, with single sign-on if you use it; systems use a platform API key. CAP AI, the Smart CR workspace and procedure-grounded screening are each switched on per deployment.

Book a demo

See a report written and screened.

We'll show a condition report written in plain words and checked as it's typed, screened with its reasons and sources, and followed through its record and trends, then answer your IT team's questions about deployment and sign-in.

  • A live walkthrough on synthetic condition reports
  • Your questions on deployment, sign-in and data
  • Where CAP AI would give your screeners the most time back

Nuclearn CAP AI is backed by Nuclearn, whose products serve over 70 facilities in North America and the U.K.

Book a demo

Tell us a little about your site and we'll set up a time.

Prefer email? contact@nuclearn.ai