Selling Outcomes, Not Tools — A Question We've Been Living With for Four Years

A recent article in Forbes Ai Native Agencies Sell Outcomes Not Software and Investors Are Paying Attention

The piece draws on Julien Bek's March 2026 essay "Services: The New Software," published by Sequoia Capital. The core argument: for every dollar spent on software, six dollars are spent on services — and AI is now capable enough to capture both. Y Combinator reinforced the trend by designating "AI-native agencies" as a priority category for its Spring 2026 batch.

Reading this, I had a simple reaction: what we've been doing in Fukui prefecture for the past four years is exactly this thesis, put into practice. And yet, the conversation seems to be happening almost entirely within a Silicon Valley frame. On the ground in Japan's SMB market, this structural shift is more urgent, more concrete, and more constrained than the venture narrative suggests.

We know where copilots don't work

At the heart of the Forbes article is Sequoia's framework: copilots sell tools to professionals who remain responsible for the output. Autopilots sell the output itself. Harvey sells legal AI to law firms (copilot). Crosby sells reviewed contracts to companies that need them (autopilot).

But there's an assumption embedded in the copilot model that often goes unexamined: it presumes there's a qualified professional inside the organization to operate the tool.

For Japanese SMBs — businesses with 5 to 50 employees — that assumption frequently doesn't hold. A hotel with no dedicated marketing staff. A tourism board whose officers have never worked with data analytics. A restaurant owner who manages their social media at night, after closing. Hand these businesses an AI tool, and there's nobody to use it.

"You can do anything if you write the right prompt" — but if you don't know what to ask, the tool is inert. This isn't a skills gap. It's a structural gap. Business operators consumed by daily operations have no bandwidth to learn a new interface, however powerful.

We've encountered this reality hundreds of times through our work with the Fukui Prefecture Tourism DX Consortium.

The numbers speak for themselves

Our work in Fukui has been ongoing since 2022 — a region-wide DX initiative centered on the Fukui Prefecture Tourism Federation, spanning local banks, newspaper publishers, tourism boards, DMOs, and hospitality operators across the prefecture. At the 2024 G7, it was selected as the only Japanese case study presented on tourism DX.

Building on that foundation, we deployed our platform — mitsumonoAI — as part of the "FY2025 Tourism DX Regional Revitalization Model Project." Approximately 160 government bodies and tourism operators used the service in their daily workflows.

The measured results:

Monthly reports that used to take facility managers an entire day to compile now generate in minutes through AI-driven data integration. Social media assets and flyer drafts that required a full week of work are now ready in 30 minutes.

"Honestly, I didn't even know where to start."

That was how one third-sector enterprise manager described the situation before deployment. Drowning in research tasks, spending full days without making tangible progress. What this person needed wasn't AI software — it was analysis delivered to their desk, and content ready to publish.

Selling outcomes is harder than selling tools

That said, the outcome-based model doesn't materialize simply by embedding AI into a workflow. In many respects, it's harder than selling tools — because what the customer is buying is not a feature, but a business result.

Not that a report gets auto-generated, but that it enables the next decision. Not that an SNS draft appears, but that the operator can confidently keep posting. Not that the AI answers a question, but that the business moves forward.

That requires domain understanding, not just technology.

What tourism operators truly struggle with isn't "no data" — it's "data I can't act on." What government bodies and DMOs face isn't just "we can't make a report" — it's "we can't get everyone on the same page." Delivering outcomes means connecting AI's output to the operational context in which it will be used.

This is where human involvement is still essential. During initial deployment, someone needs to design which tasks to replace, who reviews the results, and how often output is shared. Not full automation — but a staged approach: humans define the outcomes, AI executes, and humans embed the results into daily practice.

What four years have taught us is that an autopilot doesn't need to be fully autonomous from day one. What matters is whether, from the customer's perspective, they are buying "a tool to operate" or "an outcome that arrives." That distinction looks small but is decisive for the business model.

Start where outsourcing already exists

The point in the Forbes article that resonated most was Sequoia's advice: start with tasks that are already outsourced. If a function is already contracted out, the budget line exists, the scope is defined, and the buyer is already purchasing outcomes. Replacing an outsourcing contract with an AI-native provider is a vendor switch. Replacing internal staff is an org restructuring. The former generates far less friction.

Our tourism DX business sits squarely on this structure.

For regional tourism operators, data analysis and content creation were either outsourced to external designers and consultants — or simply left undone because the budget didn't exist. We entered not with "adopt this AI tool" but with "we'll deliver analysis and content to you." It's a replacement of existing outsourcing spend, or the first-ever provision of services that operators previously couldn't afford — for them, it's "we can finally do what we always needed but couldn't."

The article highlights Crosby offering NDA and MSA reviews at a flat ~$400 per document regardless of attorney hours, and WithCoverage purchasing commercial insurance on behalf of CFOs. Different domains, same structure: charge a fixed fee for the outcome, and use AI to drive down the cost of delivery. In our case, the outcome is analytics reports and content for tourism operators.

Declining marginal costs and compounding knowledge

The article's observation — that marginal costs for additional contract reviews or insurance placements approach zero — maps directly to our model.

As our customer base grows, so does our library of industry-specific templates and workflows. Review analysis templates for hotels. Menu development workflows for restaurants. Monthly report formats for tourism boards. Once built, these apply almost unchanged to the next customer in the same vertical. By the 100th customer, the majority of their challenges correspond to patterns already solved across the prior 99.

As the Forbes article notes, referencing Sequoia's essay: even when the product is a service, the margin profile can resemble software. A cost structure where incremental customer acquisition gets cheaper over time is fundamentally different from traditional services, where revenue scales linearly with headcount.

Furthermore, as foundational LLMs improve, template execution quality rises and human support overhead falls. Model advances translate directly into margin improvement. The evolution of competing technology doesn't work against us — a structural position we've designed intentionally.

Tool budget or work budget?

The question posed near the end of the Forbes article goes straight to the core of our business strategy: are you competing for tool budgets or work budgets? In most professional verticals, work budgets are six times larger than tool budgets.

Selling a SaaS license at €49–€500/month versus providing a service at few hundred Euro month where "analysis reports arrive weekly, SNS content is auto-generated, and the information needed for business decisions shows up organized and ready." The former competes against software budgets. The latter competes against outsourcing and labor costs. The addressable market is fundamentally different.

We offer both: an entry-level SaaS plan and outcome-oriented Partner plans. As customers grow, they naturally transition from tool usage to outcome delivery. This isn't theory — it's a migration pattern we've observed in the field.

Closing

When Sequoia suggests that the next trillion-dollar company may be "a software company disguised as a services firm," it sounds like a grand vision. But the starting point is remarkably simple.

Did the business owner in front of you achieve a better outcome today than yesterday? Can they now spend the two hours they used to burn on report creation on what actually matters — talking to customers, designing new experiences?

Globally, companies like Crosby and WithCoverage are establishing outcome-based models in legal and insurance verticals. In Q1 2026, AI startups captured 80% of global venture investment. The concentration on infrastructure — OpenAI, Anthropic, xAI — may be obscuring the more durable opportunity at the application layer: not who builds the best model, but who deploys AI to deliver the best outcomes in specific professional verticals.

The structure we've uncovered in Fukui is not a special case unique to rural Japan. Talent shortages, the absence of in-house specialists, and dependence on outsourcing budgets are universal conditions across SMB markets — including those in Europe. We're in the middle of proving this structural shift across Japan and, increasingly, across European SMB markets starting from Eastern Europe. There's still much to demonstrate. But four years in the field have given us one conviction: don't hand them a tool — deliver the outcome. In the markets we serve, this model works.