Twenty years of building on NetSuite. Now I build with AI.

The question every technical evaluator asks is how do I know it still works? Most answers to that are a reassurance. This one is an artifact.

A test suite derived from the original 1.0 source is the only kind that can catch what a conversion got wrong. Tests written from converted code encode the mistake and then pass — worse than no tests at all, because they retire the suspicion that would have prompted a manual check.

The suite ships with the project. It runs in your environment with npm install && npm test, depends on nothing of mine, and re-runs against your customizations at every NetSuite release.

Passing does not mean correct. It means the conversion preserved the behaviors observable in your 1.0 source — not that it is right, and not that anything the suite lists as uncovered was checked. Every gap is named, in plain terms: "the status codes are account-specific internal ids and cannot be verified here" is more use to you than silence. A suite that looks complete and isn't is the most dangerous thing I could hand you.

AI proposes. A person approves. Conversions are proposals — nothing is written back to your NetSuite account, and nothing reaches your repository without your named approver accepting it.

Plenty of people will tell you it does. Oracle's documentation says 1.0 "continues to be supported" but is "no longer being updated" — frozen, not deprecated, with no announced end date. I'd rather you heard that from me than found it out later and wondered what else I'd oversold.

What does have dates, published by Oracle: Token-Based Authentication stops accepting new integrations at 2027.1 and support ends at 2028.1. SOAP web services endpoints are disabled entirely at 2028.2. A great many 1.0 RESTlets authenticate with TBA, so those scripts sit inside a blast radius that is already scheduled.

Remediating integrations isn't something I offer yet — it's on the list to build, and if it's what you actually need, say so. What people ask for is how I decide what gets built next. I've written the timeline up either way, because somebody should.

The case for moving your SuiteScript is narrower and stands on its own: frozen code decaying under a platform that keeps moving, in an account where nobody can currently tell you what's in there.

The full timeline, with sources →

The migration tool exists because a NetSuite problem needed software rather than an opinion. So do these.

• Consignment order-to-cash — ownership transfers at invoicing, so the whole cycle needed tracking NetSuite doesn't do natively.
• Guided quoting — complex tier and SKU rules were producing expensive order errors, so the rules moved into the quote.
• Parent/child pricing sync — keeping child pricing aligned with a parent account by hand doesn't scale.
• Requirements and fit-gap — BRDs seeded from a 9,700-line leading-practice catalogue, with every requirement sorted into standard, licensed, customization or gap.

Every company has a list of problems it decided not to solve — not because the problem was unclear, but because the quote was six figures and living with the spreadsheet was cheaper. That calculation was correct. It is also out of date. When identity, deployment, testing and the rest already work before anyone writes a feature, what's left is the part that's actually yours — and problems that sat just under the old threshold are above the new one.

It doesn't have to be another system to log into, either. An application can be built as a SuiteApp that lives inside NetSuite, on your data, under your roles, in the navigation your people already use.

I'm Dave Weiss. I've done nothing but NetSuite work for twenty years.

That matters here more than it usually would, because the question underneath "can I trust an AI to migrate my production scripts?" isn't really about the AI. You're not trusting a model. You're trusting someone who has spent two decades in this system, who built the tool, decides what it does, reviews what it produces, and tells you plainly what it couldn't verify.

The AI-forward vendors in this market mostly don't know NetSuite. The people who really know NetSuite mostly aren't building with AI. I'd rather be the awkward third thing.

You should also know what happens after our engagement ends because it's the sensible question to ask about this kind of work: none of it depends on me. The work lands in your repository — 2.1 sources, the 1.0 originals, the regression suite, a deployable project, and every approval in the commit history. If you don't have a repository, stand one up and I’ll transfer it to you at the end. Either way you finish owning the lot, and it keeps working without me.