Sequoia Capital says the next trillion-dollar company won't sell AI tools — it'll sell AI work. We've been doing that from day one.
Last week, Sequoia Capital published an article called Services: The New Software. It's one of the most important pieces of thinking about AI I've read this year — and it describes exactly what we're building at Agenti NZ.
The core argument is deceptively simple: for every dollar spent on software, six are spent on services. The next generation of great companies won't sell you a better tool. They'll just do the work.
If you run a small business in New Zealand, this matters more than you think.
Copilots vs Autopilots
Sequoia draws a line between two types of AI companies.
Copilots sell tools. They make a professional more productive — think of an AI assistant that helps an accountant work faster, or a writing tool that helps a marketer draft emails. The professional is still doing the work. The AI just helps.
Autopilots sell work. They deliver the outcome directly to the business owner. You don't hire a faster bookkeeper — your books just get closed. You don't get a better quoting tool — your quotes just go out.
This distinction changes everything about who the customer is and what they're paying for.
A copilot sells to the accountant. An autopilot sells to the business owner who needs their accounts done. A copilot competes on the software budget — maybe a few hundred dollars a month for a subscription. An autopilot competes on the services budget — the tens of thousands you spend on outsourced professionals every year.
The services budget is where the real money is.
Why this matters for Kiwi businesses
Here in New Zealand, small businesses spend enormous amounts on outsourced services. Bookkeeping. Payroll. IT support. Marketing. Recruitment. Insurance broking. Legal compliance. For most SMEs, these costs add up to far more than what they spend on software subscriptions.
Sequoia's argument is that AI is now good enough to deliver many of these outcomes directly — without the middleman. Not by giving you a better tool to do the work yourself, but by actually doing the work.
This is exactly what we built Agenti NZ to do.
When we deployed AI agents for EVX, an EV rental business, we didn't give them a better booking tool. We built agents that handle the bookings. What used to take 45 minutes per reservation now takes 5. The business didn't learn new software — they just got the outcome.
When we worked with Kiwi Outdoor Gear, we didn't sell them an inventory dashboard. We deployed agents that manage the inventory — tracking stock levels, predicting demand, adjusting prices, and reordering automatically. Stockouts dropped 85%. They didn't need to become data analysts. The work just happened.
That's the difference between a copilot and an autopilot.
Intelligence vs Judgement: a framework for every business owner
The most useful idea in Sequoia's piece is the distinction between intelligence work and judgement work.
Intelligence Work
Rules-based and repeatable. It's complex, but it follows patterns. Translating a quote request into a formatted PDF. Matching an invoice to a purchase order. Categorising expenses for GST. Screening CVs against a job description. Sending follow-up emails on a schedule.
Judgement Work
Requires experience, instinct, and taste built over years. Deciding which client to prioritise. Choosing whether to take on a risky project. Knowing when to push back on a supplier. Reading the room in a sales meeting.
AI has crossed the threshold where it can handle most intelligence work autonomously. Judgement work is still yours — and that's exactly where your time should go.
The problem is that most business owners spend 60–70% of their day on intelligence work. The invoicing, the scheduling, the data entry, the follow-ups, the reporting. That's the work AI agents can take over today.
At Agenti NZ, we've been saying this in simpler terms: automate the boring, focus on growing. Turns out Sequoia agrees — they just needed a few more paragraphs to say it.
The outsourcing wedge
Sequoia's playbook for AI autopilots starts with a specific insight: begin where work is already outsourced.
Why? Three reasons. First, the business has already accepted that this work can be done externally. There's no mindset shift required. Second, there's an existing budget line — you're already paying someone to do this. Third, the buyer is purchasing an outcome, not managing a process.
Replacing an outsourcing contract with an AI agent is a vendor swap. Replacing an employee is a reorganisation. One is a purchasing decision. The other is a change management project.
For New Zealand SMEs, the outsourced services that are ripe for this shift include:
Bookkeeping and accounting
New Zealand, like the rest of the world, has an accountant shortage. Most SMEs outsource their books to a bookkeeper or accounting firm. The work — categorising transactions, reconciling accounts, preparing GST returns — is almost entirely intelligence work. An AI agent connected to Xero can do most of it continuously, not just at month-end.
IT support
Every small business outsources IT to some degree. Patching, monitoring, user setup, password resets, email issues — it's repetitive intelligence work running across thousands of similar environments. An AI agent can handle the first line of support and escalate only when genuine judgement is needed.
Recruitment
Writing job ads, screening applications, scheduling interviews, sending rejection emails. The top of the hiring funnel is pure intelligence work. The judgement — assessing culture fit, making the final call — stays with you. But the hours of admin don't have to.
Marketing and customer follow-up
Sending quotes, following up on leads, posting to social media, responding to enquiries. These are the tasks that fall through the cracks when you're busy — and they're the ones AI agents handle brilliantly.
What this means for our approach
Reading Sequoia's piece confirmed something we've believed from the start: the value isn't in the tool, it's in the outcome.
When a tradie gets a quote out in 5 minutes instead of 45, that's not a software feature — that's revenue recovered. When a retailer's inventory reorders itself before a stockout, that's not a dashboard — that's sales protected. When a professional services firm's proposals go out the same day instead of next week, that's not productivity — that's deals won.
We don't sell AI tools. We build AI agents that do the work. You own the system, you own the code, you own the data — but the outcome is what you're paying for.
The opportunity for New Zealand
There's something in Sequoia's piece that's particularly relevant here. They note that the highest-opportunity categories are where the work is primarily intelligence-based, already outsourced, and served by a fragmented industry of small providers.
That describes half the professional services industry in New Zealand.
We're a country of small businesses — 97% of NZ businesses have fewer than 20 employees. Most of them outsource key functions to other small businesses: their accountant, their IT person, their bookkeeper, their marketing freelancer. The services spend is enormous relative to revenue.
AI autopilots don't just help the business that adopts them. They reshape the economics of entire service categories. When a $500/month AI agent can do the work of a $2,000/month outsourced service, every business in the country benefits.
That's not a threat to the professionals doing this work — it's a signal that their role shifts toward judgement. The best accountants aren't doing data entry; they're advising on strategy. The best marketers aren't scheduling posts; they're crafting brand stories. AI handles the intelligence work so professionals can charge for what actually matters: their judgement.
Where we go from here
We're building Agenti NZ around this thesis. Every AI agent we deploy is designed to deliver a business outcome, not a software feature. Our three-tier model reflects this:
- •The AI Discovery Audit identifies which of your tasks are intelligence work (automate now) and which are judgement work (keep doing yourself). It maps the opportunity.
- •The Custom Agent Build creates the autopilot — agents that handle the intelligence work end-to-end, integrated with the tools you already use.
- •The Support & Optimisation retainer keeps the agents improving. Because unlike an outsourced contractor who does the same thing every month, an AI agent that learns from your business gets better over time.
Sequoia wrote their piece for Silicon Valley founders raising venture capital. But the underlying insight belongs to every business owner: AI is good enough to do the work now. The question isn't whether to adopt it — it's whether you'll be the one selling AI-powered outcomes, or the one still paying full price for intelligence work a machine can do.
If you're a Kiwi business owner wondering where to start, that's what our Discovery Audit is for. We'll map your intelligence work, show you the ROI, and build the agents that do it.
Because the future isn't better tools. It's better outcomes.
We're Agenti NZ — we build autonomous AI agents for Kiwi businesses. If you want to explore what AI autopilots could look like for your business, reach out at [email protected].