Stop Selling Software. Start Selling Outcomes: Inside the AI Services Shift

Why the next wave of value won’t come from selling tools, but from selling AI-delivered insights and outcomes.

When software stops being the product

I’ve been struck over the past year by what feels like a tectonic shift in the software business: AI is quietly turning “software as a service” into “service as software,” and pushing us from selling seats and features to selling outcomes and insights as the real unit of value. What used to be a tool that helped humans do the work is rapidly becoming an intelligent layer that does the work, or at least a meaningful chunk of it, which means the only pricing and business models that make sense are those tied to the value created, not to how many people happen to be clicking around in a UI. In my view, this is not a nuance or a side‑show; it’s a structural change in how value will be captured in software, and every serious software company now faces an uncomfortable but urgent question: how fast can we move from “butts in seats” to outcome- and insight-based models before our own AI makes our legacy pricing look absurd

“In an AI‑first world, the real product is no longer software licenses—it’s AI‑delivered outcomes and the insights that drive better decisions.

This wave is clearly demonstrated by a new generation of companies who are taking direct accountability for outcomes—claims paid, legal documents drafted, revenue collected, fraud caught—and using AI as the engine that does the bulk of the work. Their customers are not buying tools or access. They are buying a result and are increasingly indifferent to whether that result was produced by a human, an agent, or a hybrid.

Emergence Capital has spent the last two years studying this model, which they call AI-Native Services, and codified the patterns in their AI-Native Services Playbook and companion essay, “Why AI-Native Services, and Why Now.”

• Article: Why AI-Native Services, and Why Now

Sequoia, in parallel, has argued that “services are the new software” and that the next trillion-dollar company will not sell software as a product but outcomes, using AI-powered software and human expertise under the hood. This piece builds on those theses and asks a practical question: if you’re already running a SaaS or data platform, how do you actually pivot toward these AI-forward business models, pricing schemes, and value-delivery patterns?

• Article: services are the new software

From SaaS to AI services: what actually changed

The SaaS era was about capturing the software budget. Vendors sold tools that helped humans work faster or better, and the economic unit was the seat, license, or feature bundle. AI-native services aim at a bigger pool of spend: the labor and services budget that dwarfs pure software in most enterprises. For every dollar a typical enterprise spends on software, it spends a multiple of that on services and people doing the work.

Sequoia’s “Services: The New Software” thesis puts it plainly: the next wave of value will come from companies that look like service providers on the outside but are, economically, AI-native factories for outcomes and insights on the inside. Emergence’s playbook complements that by showing how these companies actually operate: they collapse software and services into one integrated system, own the full stack of work, and use AI to deliver 5–10x improvements in speed or throughput while still operating at software-like gross margins.

At a structural level, this shift changes the unit of value. Traditional SaaS sells tools and usage; AI-native services sell resolved tickets, adjudicated claims, cleared backlogs, validated transactions, or decisions made with better signal. The buyer is already purchasing an outcome today, whether via internal teams or outsourced providers; replacing that contract with an AI-native service is a supplier swap, not a new line item.

Insights as the new business model

Most analytics and BI tools stopped at dashboards; they generated charts and reports, and humans turned those into decisions. AI-native services go one step further: they ingest raw, messy data and output a decision, an action, or a fully prepared work product. In that world, the monetizable unit is no longer the dashboard—it is the insight plus the action it enables.

You can see this in verticals Emergence highlights: insurance services that use AI to process claims and detect fraud; fund administration firms that run an end‑to‑end process on behalf of clients; revenue cycle providers that use AI agents to work denials and recover dollars. In each case, the provider’s moat is the compounding “insight flywheel”: every engagement generates labeled outcomes and operational data that improve model performance, prediction quality, and unit economics over time.

“If your revenue model still starts with seats and feature bundles, you’re probably under‑monetizing your AI; if it starts with outcomes, you’re on the right side of the shift.”

This is the core of “insights as the business model.” As AI takes over more of the execution, what you truly own is a system of intelligence about a domain: how often certain patterns lead to specific outcomes, what interventions change those outcomes, and how to package that into a contract your customers understand. The more work you do, the better your insights become; the better your insights, the more attractive your pricing and SLA can be, and the harder it becomes for a new entrant to catch up.

The economics of selling outcomes

Once AI starts doing real work, traditional pricing breaks. Per-seat models actively punish you for being successful: if your AI allows a customer to reduce headcount by 50–80%, their seat count drops along with your ARR. This is why both Emergence and a growing body of pricing research argue that AI-native services are the natural home for outcome-based models.

“The difference between a glorified services firm and an AI‑native service is simple: in one, people do the work with AI as a tool; in the other, AI does the work and people teach it to get better.”

Outcome-based pricing charges only when a specified result occurs—tickets resolved, cases closed, dollars recovered, qualified leads generated—rather than for attempts or usage. Intercom’s Fin and similar AI support agents, for example, charge per resolved conversation rather than per user, transferring performance risk to the vendor but aligning cost directly with value delivered. Emergence’s playbook generalizes this: when you own the outcome, attribution is clean, and the service itself is the outcome, making AINS the most natural home for outcome-based economics in the AI stack.

For investors, this changes what “good” looks like. Growth rate and logo retention aren’t enough; you have to see AI leverage in the financials and operations. Emergence warns about “Mirage PMF”—situations where revenue growth is real, but AI isn’t doing the work and margins aren’t expanding. Leading indicators include a rising share of work done by AI, falling human review time per unit of output, and improving revenue per employee; lagging indicators show up in honest gross margins that include inference and human-in-the-loop costs in COGS.

How an existing SaaS company actually makes this pivot

You don’t need to be a greenfield AI startup to move in this direction. Over the past year, I’ve watched an established vertical software and data company work through a deliberate shift from “selling tools and access” to “selling outcomes and insights,” using Emergence’s AI-Native Services Playbook as a compass rather than a script. The steps they took are instructive for any incumbent thinking about a similar move.

1. Name the outcome and admit you’re already a service

The leadership team started by changing how they described themselves. Instead of presenting as “a platform with seats,” they reframed the business as “an outcome engine” that delivers a specific result: clean data, adjudicated items, qualified opportunities, or similar. In reality, their customers already cared about those results, not log‑ins or BI widgets. Adopting the AI-native vocabulary made it easier to plug into the frameworks Emergence and Sequoia had put into the ecosystem and to articulate their differentiation to investors and senior hires.

2. Diagnose Mirage PMF and pick a single AI leverage metric

They then pressure‑tested whether they were getting true AI leverage or just scaling by throwing people at demand. The question was simple: as volume grows, is the cost to deliver growing sub‑linearly or roughly in lockstep? To answer it, they borrowed Emergence’s idea of a single north-star productization metric—an internal “HURT”‑style measure of how much human work remains after AI and the platform have done their part. Candidates included human review minutes per unit, percentage of outputs produced end‑to‑end with zero human touch, or cost per structured field at a given quality threshold. That one number became the litmus test for whether they were building an AI-native service or just a services firm with AI in the toolchain.

3. Treat delivery as the product and turn bespoke work into roadmap

On paper, everything flowed through the platform; in practice, many key accounts depended on “humans plus platform” workflows to fill gaps. Instead of treating this as a permanent operating condition, they treated every human‑heavy step as input to a product backlog. Any process that required manual effort at scale became a roadmap item for automation or standardization, with explicit owners and timelines.

This matched Emergence’s emphasis on “sleeping at your customer’s office”: founders and senior ICs stayed close to real work, not to grow services headcount but to observe patterns that could be codified and handed off to AI and software. Over time, that shifted the center of gravity from bespoke projects to an integrated factory that learned from every engagement.

4. Hire product to sit between engineering and delivery—earlier than feels comfortable

Counterintuitively, they invested in a strong product leader for the “applications” or “solutions” layer even though end customers never logged into that software directly. Emergence notes that many AINS founders delay this hire and pay for it later when the roadmap is driven by whoever is shouting loudest. In this case, the product leader’s mandate was clear: listen deeply to delivery, but say “no” to bespoke work that didn’t advance productization or move the AI leverage metric.

That bridge role created a disciplined feedback loop between the people doing the work and the people building the AI, while protecting against the failure mode of becoming a traditional services shop responding to endless one‑off requests.

5. Fix GTM: “it’s the demo, stupid”

Go‑to‑market was still “SaaS‑shaped”: lots of talk about coverage, accuracy, lift, and benchmarks, very little visual proof of the AI doing the work. Following Emergence’s advice, they built a real demo environment that let prospects watch raw input come in, flow through AI and normalization layers, and emerge as the outcome they actually pay for.

The buyer would never touch this interface in production, but watching the “factory” operate did two things: it halved sales cycles in some segments and made the differentiation versus legacy manual or semi‑manual providers obvious. It also forced the team to confront where the process was still too human‑heavy, feeding back into the productization roadmap.

6. Design a bridge from labor‑based to outcome‑based pricing

Rather than flipping a switch to pure outcome pricing, they treated pricing as a series of controlled experiments. By segment, they defined the eventual unit of value the customer would gladly pay for—per record processed, per checklist completed, per case closed, per dollar recovered—and then designed a bridge from current volume or labor‑based models to that target.

This echoed Emergence’s pragmatic guidance: start with market norms while your AI is maturing, but set explicit timelines and milestones for migrating to outcome-based pricing where incentives and economics truly align. Some segments moved quickly, where outcomes were easily measurable and repeatable; others required more time to reduce variance and build trust.

7. Turn the data flywheel into a board‑level asset

Finally, they elevated the “data flywheel” from a technical talking point to a governance priority. In AI-native services, the data generated by doing the work—inputs, intermediate states, labels, and outcomes—is the core product advantage. Leadership systematically reviewed MSAs, channel agreements, and registry or platform terms to ensure they had the right to use service data to improve the service.

They also asked a tough question of each partnership: “Are we still the system of record for what happens, or have we become a sub‑supplier inside someone else’s relationship?” Emergence is explicit that if you lose the customer relationship, you lose the data flywheel; this company made maintaining that direct relationship a non‑negotiable design constraint for GTM and channel strategy.

Exits, valuations, and who buys AI-native services

This shift doesn’t just change how you operate; it changes how you’re valued and who is likely to buy you. The last 18 months have seen a painful repricing in public SaaS as investors reassess long‑term growth and worry about AI commoditizing software features. At the same time, AI-native companies with strong AI leverage and outcome-based models are achieving faster paths to scale and, in many cases, richer multiples.

For AI-native services, the right analog is not a classic low-margin body‑shop. When AI is doing a material share of the work, revenue per employee rises and honest gross margins (with inference and human‑in‑the‑loop in COGS) look much more like high‑quality software than traditional services. Add a durable data and insight moat—proprietary outcome data, model performance advantages, long‑lived contracts—and you get something closer to an owned system of intelligence in a vertical than a generic outsourcing vendor.

“AI-native services are structurally set up to capture the full upside of the disruption they unleash—owning the data flywheel, the decision logic, and the economics of the work itself.”

That shows up in exits. Strategic software buyers are starting to look for AI-native services they can bolt onto existing platforms to defend their base and convert “tool” relationships into “outsourced outcome” relationships. BPO and IT services consolidators see AI-native acquisitions as a way to re‑rate legacy books of business: inject AI into traditional engagements, expand margins, and sell the combined entity at a higher multiple. Vertical incumbents—banks, insurers, healthcare systems, industrials—are exploring acquisitions of AI-native services that sit too close to core operations to leave in third-party hands.

For founders and boards, the message is simple. If you look like a body‑shop with some AI sprinkled in, you will be valued on services multiples. If you look like an AI‑leveraged outcome machine with a protected data flywheel and clear, outcome‑aligned pricing, you have a credible argument for software‑like, or even premium, multiples. The decision to sell tools vs outcomes doesn’t just shape go‑to‑market; it drops you into a different bucket in the acquirer’s valuation model.

Stop selling tools. Own the outcome.

As foundation models become capable of doing more of the work humans used to do, seat‑based SaaS economics will come under increasing pressure. The companies that thrive will be those that take direct accountability for outcomes, use AI as the engine to deliver them, and build compounding insight moats from every engagement.

Emergence’s AI-Native Services Playbook and Sequoia’s “Services: The New Software” essay provide a shared vocabulary and a set of pressure tests: do you have real AI leverage, or Mirage PMF? Is your pricing aligned with outcomes? Is your data flywheel strengthening with every customer? The good news is that many SaaS and data companies are already closer to AI-native services than they think. The hard news is that making the shift requires deliberate choices—in product, GTM, pricing, contracts, and even how you talk about what you do.

If your revenue model still starts with seats and feature bundles, you are probably under‑monetizing your AI. If it starts with outcomes, and you can measure and improve how much work AI actually does, you’re on the right side of the shift.

Conclusion

In the end, this wave isn’t just about better tools or clever pricing hacks; it is about finally aligning how we charge for software with where AI actually creates value, and that is where the returns get interesting for everyone involved. AI-native, outcome- and insight-driven services are structurally set up to capture the full upside of the disruption they unleash—owning the data flywheel, the decision logic, and the economics of the work itself—and that combination is tailor-made for outsized value creation. For founders, it has never been a better time to build: small, focused teams can stand up AI-native services in narrow verticals and scale to meaningful revenue with a fraction of the capital it used to take. For investors, it has rarely been a better time to underwrite durable advantage: if you can separate Mirage PMF from real AI leverage and back the companies that truly sell outcomes, not seats, you have a chance to participate in one of the most significant value shifts the software industry has ever see

Appendix: background reading and references

• Emergence Capital – “The AI-Native Services Playbook” (Spring 2026): Tactical guide to building AI-native services across team, PMF vs Mirage PMF, delivery, roadmap, GTM, pricing, defensibility, metrics, and M&A.

• Emergence Capital – “Why AI-Native Services, and Why Now” (2026): Strategic framing of why AI-native services are emerging and how they structurally differ from SaaS.

• Emergence Capital – “AI-Native Services: The Definitive Guide” hub: Updated set of essays, portfolio examples, and operational frameworks for AINS founders.

• Sequoia Capital – “Services: The New Software” (March 2026), by Julien Bek: Thesis that the next trillion‑dollar company will sell AI‑delivered outcomes rather than software, and that AI-native service firms will capture the much larger services and labor budgets.

• Supplemental – AI pricing and outcome‑based models: a16z and other investors’ AI pricing playbooks, plus case studies of per‑resolution and outcome-based AI pricing in customer support and other verticals