The NetSuite Cost Curve

My consultancy is rebuilding how we deliver NetSuite work — around AI.

Here's the bet: a NetSuite customization that took six weeks last year should take two by the end of this year. The work isn't getting easier; the lifecycle is getting compressed at every stage.

We're early in this. But the framework is clear, and so is where it's actually happening.

Phase 1 — Discovery: from scattered artifacts to a real spec

Traditional NetSuite discovery looks like this: three stakeholder meetings, a Slack thread, two PDF requirements docs, a folder of screenshots, and a spreadsheet of edge cases. Then a senior consultant spends a week trying to reconcile it all into a coherent specification.

With AI, all of that becomes a single context window.

Meeting transcripts, Slack exports, requirement docs, screenshots, sample data — fed into one place, with the right structure and prompting. You get:

• A first-draft technical specification in hours, not days
• Surfaced contradictions between what stakeholders said across different meetings
• Gaps and ambiguities flagged that the team didn't notice
• A document everyone can react to early, when changes are still cheap

The senior consultant still owns the spec. They're just editing a strong draft instead of writing from a blank page.

Phase 2 — Build: SuiteScript generation with guardrails

Once the spec is locked, AI generates scaffolded SuiteScript 2.x — user events, scheduled scripts, RESTlets, Suitelets — directly from the requirements.

What used to be days of boilerplate becomes hours of review and refinement.

The senior engineer still owns architecture, governance, and final accountability. The model writes the first 60–70%. The human writes the 30% that actually matters: the edge cases, the integration logic, the things AI doesn't know about your specific account.

It's not "AI codes for you." It's "AI handles the parts of coding that were never the hard part."

Phase 3 — Testing: the part everyone skips

This is the phase I'm most excited about.

Most NetSuite customizations ship with manual sandbox testing — a few clicks, a thumbs-up, deploy. Real test coverage is rare because it's expensive.

AI changes that math entirely.

Given a spec and the code, AI can generate:

• Test plans covering happy paths and realistic edge cases
• Synthetic test data sets across record types
• Automated validation that exercises the customization end to end
• Regression catches before anything reaches production

And on the horizon: browser-driving AI agents that can navigate a NetSuite sandbox the way a human tester would — open a sales order, fill the fields, save the record, verify the workflow fired, check the GL impact. No test harness. No API plumbing.

The tools are still early. Claude's Chrome extension and similar browser agents aren't yet reliable enough for unsupervised QA against complex NetSuite workflows. But the trajectory is unmistakable — and when this matures, comprehensive UI-level test coverage on small and mid-sized NetSuite projects becomes economically trivial. We're tracking the space closely.

For the first time, comprehensive testing on small and mid-sized NetSuite projects is economically viable. Not because the technology is new — but because the cost dropped enough that "good QA" can finally be the default, not the exception.

What stays expensive

AI commoditizes configuration and boilerplate. What gets more valuable is judgment — knowing what to build, where the gates and exception handlers need to live, when a stakeholder is asking for the wrong thing, which customization will become technical debt in eighteen months.

That's what senior NetSuite experience is for. And it's the part of the work that isn't getting cheaper.

Quite the opposite — when execution gets cheap, the cost of building the wrong thing gets cheap too. Bad decisions ship faster now.

Which is also why the platform itself — its accounting integrity, its transaction model, the safeguards it brings out of the box — matters more in this world, not less. AI gets faster at scaffolding. The foundation underneath still has to be right.

The compounding effect

Each phase individually is maybe 50–70% faster. That alone is significant.

But here's what most people miss: when discovery compresses, build starts sooner. When build compresses, more time is left for real testing. When testing compresses, fewer bugs reach production — which means less rework, which means the next project starts sooner.

The result isn't a linear speedup. It's a system that produces better customizations in a fraction of the time, with QA standards that used to be impossible to afford.

For most companies running NetSuite, the math just shifted.

The customization you've been quoting at $40K and six weeks should be closer to $15K and two weeks. The improvements sitting on your "someday" backlog — too expensive, not strategic enough — are back on the table. The integrations you've been told would take a quarter? Closer to a few weeks. Test coverage stops being a luxury reserved for big-budget engagements.

That's the actual story of AI in NetSuite work — not job replacement, not magic, not buzzwords. Just a dramatic reduction in the cost of getting NetSuite to do what your business actually needs.

The question for anyone running a NetSuite shop is no longer "can NetSuite do this?" It's "why is my consultancy still pricing like it's 2023?"

I'll be wrong about parts of this — the playbook is still being written. But the cost structure is moving in one direction, and ignoring it isn't a strategy.

There are deeper questions this post doesn't try to answer — whether AI ultimately erodes NetSuite's value entirely, what the consulting business model looks like in five years, where NetSuite's real competitive advantage actually lives. Those deserve their own conversation. The operational question above is the one that matters this quarter.

What's the most expensive phase of your NetSuite projects right now? I'm curious where teams are still feeling the most friction.