Every script, workflow, and saved search in a NetSuite account was put there by someone, for a reason. A few years in, half of those people have moved on, most of the reasons are undocumented, and the account keeps running on automation that nobody would dare to switch off because nobody knows what it does. That's the platform side of technical debt, and it's invisible from every standard screen.
Part one covered thirteen Operational Intelligence prompts new in the Sonar edition of my NetSuite AI Prompt Library, about the shape, data quality, and rhythm of an account. These twelve are about governance: what runs, who owns it, what's coming, and how the account is being used.
If you're new to this, Sonar AI is an AI agent that runs inside NetSuite. Every prompt in the library is a playbook that I engineered and tested against live NetSuite data, and you run it inside your own account, against your own records.
What Runs
NetSuite Customization Audit is written for an IT audit manager. It inventories every deployed script, active workflow, and custom record type and assesses them as IT general controls: a deployed script that touches financial transactions and isn't in a change log is a change-management risk, and the prompt flags it rather than assuming it was approved. Findings are control observations, and the prompt is explicit that they become material weaknesses only if they could produce a material misstatement.
Script and Deployment Sprawl Audit goes deeper on the scripts. It inventories every script and deployment record, quantifies technical debt, finds undeployed code, and ranks performance risk, separating signal from noise so the dashboard is something a person can act on. Workflow State Topology does the same for workflows: every active one, its trigger conditions, and its record type scope, mapped to expose gaps where nothing is automated and redundancy where two workflows fight over the same record.
Two prompts cover the reporting layer. Saved Search and Workbook Inventory lists every saved search and Analytics workbook, flags stale and duplicate assets, and surfaces governance gaps such as private silos and naming problems. Saved Search Ownership and Exposure Map, which was already in the library, answers the harder question: who owns the load-bearing searches, what breaks when that person leaves, and which searches are single points of failure. It's honest about a limit that catches people out: NetSuite exposes no reliable saved-search table to SuiteQL, so the estate is enumerated through the tools available to the current role, and private searches belonging to other users are invisible. A search counts as script-referenced only when its id appears in script source.
What Is Coming, and What Is Used
NetSuite Release Impact Review Generator takes an upcoming release, say 2026.2, and answers the only question that matters for one account: of everything changing, what could affect us, and what must we test? It reads the release notes, profiles the live account with real queries, and cross-references the two into a prioritized checklist. Every rating has to be evidence-backed. "High, because you have five custom-tool scripts, ids listed" is a finding; "this might affect you" is filler. And the absence of a table is itself a finding, because it usually means the feature isn't enabled.
NetSuite Utilization and Adoption Analyzer answers the question every NetSuite owner eventually asks: are we fully using what we pay for? It distinguishes enabled from adopted. Features tell you what's switched on; transaction and record volumes tell you what's used, and the gap between the two is the finding. Every recommendation has to cite the number that justifies it, and it delivers a prioritized improvement roadmap rather than a list of complaints.
Documents in Motion
Open Document Decay Curve plots the aging of open invoices, bills, and orders by the month they originated, so you can see whether documents from six months ago are closing at the same rate as documents from last month. A healthy business closes documents; a decaying curve says where collection or payment is systematically slow. It validates its totals against the native receivables and payables balances before publishing.
Sales Order Edit Storm finds orders that were edited an abnormal number of times after submission, which is a leading indicator of quoting instability, approval bypass, and churn risk. The prompt is careful with the conclusion: a high edit count is a signal, not a verdict, since complex orders get edited for legitimate reasons, and anything implying approval bypass or individual accountability is flagged for a person.
Order Document to Draft Sales Orders is the one prompt in this group that creates something, and it's built around that fact. It converts customer order documents, PDF text, CSV, an email paste, or a file in the file cabinet, into staged draft sales orders. It calibrates to the instance first, resolves items deterministically by SKU, then UPC, then customer part number, then exact display name, and never guesses. Every validation failure becomes an exception with a proposed disposition, and there are two hard gates: the full normalized line table for review after parsing, and a dry-run preview per batch before anything is created. Drafts are drafts, pending approval, never auto-approved, never fulfilled.
Getting Data Out, and the Factory Floor
NetSuite Warehouse Download is a structured, paginated bulk export of core tables in a shape a data warehouse or BI tool can load, complete and consistent, with the fields that SuiteQL won't return noted as gaps rather than silently dropped. Yield Rate Analyzer is for manufacturers: work order yield rates, build completion against planned quantities, scrap and adjustment activity in dollars, and quality trends, each finding tied to a specific work order, item, or process.
A Note on Ownership
Most of these prompts produce an inventory, and an inventory without an owner column is a list that gets read once. The customization audit, the sprawl audit, and the two saved-search prompts all try to attribute what they find to a person, a bundle, or a script, and they're honest when the attribution isn't in the data. A script whose owner has left the company is a different kind of finding from a script that's simply undocumented, and the reports keep the two apart.
That's also why these prompts are useful before a release rather than after one. The Release Impact Review tells you what to test. The inventories tell you who should test it, and what nobody will notice until it breaks.
Wrapping Up
The question I've come to ask about any automation in a NetSuite account is not "what does it do" but "who would notice if it stopped." If the answer is nobody, it's a candidate for retirement. If the answer is everybody, it needs an owner and a change log. These twelve prompts are how you get to either answer with evidence instead of a guess.
All twelve are in the paid tier of the Sonar AI Prompt Library, under Operational Intelligence.