Start where you can check the answer

This is how we would answer it. We were in the room for the launch, and our founder is a sitting CIO who has to answer for the systems afterwards.

What was actually announced

The launch covered more than 45 skills across six areas: financial analysis and close (explaining variances, spotting close blockers, reviewing reconciliation differences), billing and cash (billing gaps, overdue invoice priority, liquidity), inventory and fulfilment (stock imbalances, fulfilment delays, work-order cost differences), projects and services, sales and customer service, and workforce costs. NetSuite made them available to all customers through GitHub and AI service marketplaces, and says more than 13,000 customers already use its AI Connector Service, which is how most businesses will reach them.

Alongside the skills came AI portlets, AI workspaces (including an Autonomous Close workspace), and some industry-specific agents. Analyst coverage after the event was useful in separating what ships now from what is coming. Futurum's view was that the skills, portlets, workspaces and selected industry agents are generally available where NetSuite Next is offered, while the tools for building and governing agents at scale (SuiteAgent Studio, SuiteFlow Agent and the agent management workspace) are not yet. It also noted that no customer has yet published measured results on days taken out of the close.

That distinction is the most important thing in this article. The autonomy is arriving before the controls that make it safe to rely on. It does not mean you should wait. It means you should choose your starting point carefully.

The test for where to start: can someone check it in two minutes?

Every AI output in a finance process needs a reviewer, at least until it has earned trust. So the best first skills are the ones where a competent person can tell quickly whether the answer is right.

Good candidates:

• Variance explanations. A controller already knows roughly why a cost centre moved. If the skill's explanation matches, it has saved them an hour. If it does not, they will spot it immediately.
• Overdue invoice prioritisation. Credit control already has a view of who to chase. The skill's ranked list is easy to compare against it, and disagreements are informative.
• Billing gaps. Each flagged gap can be confirmed or dismissed against the contract or the order. The cost of a false positive is a few minutes.

If you want checks like these running every month, rather than only when someone thinks to ask, that is the job of our software arm, Revaion, which does it for pricing and supplier recovery inside NetSuite.

Weaker first candidates are the ones where checking the output means redoing the work. If verifying an answer takes as long as producing it, the skill has not saved anything yet; it has added a step.

Fix your roles before you switch anything on

The skills run inside the permissions people already have. That is a sensible design. It also means the AI inherits whatever state your role design is in.

In most NetSuite accounts we have seen, roles have grown by exception over the years: a permission added for a month-end task, never removed; a contractor role cloned for a new starter. None of that mattered much when the person using the role had to know where to look. An AI skill does not get lost. It will find everything the role can reach.

Before switching on skills for a team, review the roles they use. It is the same access review that starts any ISO 27001 and cyber governance programme. It is unglamorous work, it rarely takes long, and it is the difference between "the assistant only sees what you see" being reassuring and being a risk.

Give every skill an owner

The failure mode for AI in finance is not usually a dramatic error. It is drift: a skill switched on in a burst of enthusiasm, used heavily for a month, then quietly ignored, still running and still reachable.

So name an owner for each skill you enable. One person who looks at its output weekly, keeps a short note of where it was right and wrong, and decides at the end of the quarter whether it stays. That note becomes your evidence for going further, and for your auditors when they ask what the AI is doing in your close. It is how we approach AI and automation generally: contained, measured and owned.

Leave the close until the governance ships

An autonomous close is the headline everyone remembers, and it is the one we would leave until last. The close produces numbers that directors sign and auditors test. That needs audit trails, approvals and controls your auditors accept, and a finance leader willing to put their name to figures assembled partly by an agent.

The workspace for it exists now. The governance tooling does not, yet. Use the time between now and then to run the contained skills, build your evidence and get your roles in order. When the governance arrives, you will be ready to use it rather than starting from scratch. The discipline is the same one that decides most ERP implementations: decisions owned by the business, not delegated to the tool. If your reporting already depends on governed data feeds out of NetSuite, you have a head start.

Honest limits

This advice has edges, and they are worth stating.

• We have not seen your account. The sequencing above is the general case. A business with unusual processes, heavy customisation or a specific pain point might rightly start elsewhere.
• The release is new. Everything here is based on what NetSuite announced in late October 2026 and early analyst coverage. Real-world results will take months to appear, and some of this will need revisiting.
• Doing it yourself is entirely possible. Nothing here needs a consultant. What it needs is someone with the time to review roles, pick the first skills and own the output. If you have that person, you do not need outside help.

Frequently asked questions

Are NetSuite SuiteSkills available now?

Yes. NetSuite announced them on 26 October 2026 and made them available to all customers through GitHub and AI service marketplaces.

Do SuiteSkills change what data a user can see?

No. They run within existing NetSuite roles and permissions, which is why reviewing your roles first matters.

Can SuiteSkills run our month-end close?

NetSuite has an Autonomous Close workspace, but the tools for governing agents at scale are not yet generally available, and no customer has published measured close results. Treat a fully autonomous close as a roadmap item for now.

Which SuiteSkills should a mid-market business start with?

Ones whose output a person can check quickly: variance explanations, overdue invoice prioritisation and billing gaps are good candidates.

Do we need a consultant to do this?

No. You need someone with the time to own it. If that capacity is the problem, a short review can help you sequence it.