The Rise of AI-Native Services

Historically, software made individual tasks easier, but people still had to connect the pieces and complete the work. As point solutions proliferated, that meant coordinating work across more tools and systems.

AI changes the division of labor. Foundation models (like ChatGPT, Claude) can now reason across steps, retain context, and act across systems, allowing software to increasingly deliver the outcome itself. This creates a fundamentally different category: software that can capture services revenue while increasingly automating the labor required to deliver it.

Demand is pre-existing across massive markets, from legal ($400B+)² and staffing ($700B+)³ to consumer categories like home services ($650B+)⁴, where customers already delegate the work and allocate budget to the outcome. Rather than creating new markets, AI-native services can convert existing services expenditures into software revenue, creating a faster, more capital-efficient path to venture scale.

Even more compelling, AI-native services create a rare venture dynamic: scale improves both economics and defensibility. Every outcome delivered can improve the next, lower the cost of execution, and unlock more work to own. That compounding happens through a powerful flywheel: own one valuable outcome, learn through execution, and expand through trust.

The AI-Native Services Flywheel

Own One Valuable Outcome

The flywheel starts with a narrow outcome the company can own end-to-end. The narrow scope lowers the barrier to adoption, making it easier for customers to delegate the work and for the company to close the execution loop. Full ownership creates two structural advantages over traditional software: more of the economics of completing the work and direct access to the decisions that produce it.

Owning the outcome is what allows AI-native services to compete for much larger services budgets. Rather than selling tools that make the work more efficient, they sell completion itself while increasingly replacing the labor required to deliver it with software.

Ownership also puts AI-native services directly in the execution path. As decisions unfold, they capture the preferences, tradeoffs, exceptions, actions, and feedback that shape the outcome, creating a richer record of how and why the work gets done.

This is the difference between automation and ownership. Automation makes individual steps easier. Ownership completes the job and learns from every execution.

Learn Through Execution

Owning execution end-to-end means every completed outcome can improve the next, reinforcing successful paths, revealing new exceptions, and giving future decisions more context than those before them.

Closed learning loops have already proven to be a powerful foundation for building some of the world’s most valuable technology companies, from Meta ($1T+) and Netflix (~$300B) to TikTok parent ByteDance ($500B+). These platforms recommend content, observe what users watch, skip, click, or share, and use that feedback to improve the next recommendation. More usage generates more learning, which improves the product and drives greater engagement, retention, and monetization.

AI-native services apply the same dynamic to performing work. Instead of recommending an action and learning from the user’s response, they make decisions, execute the work, observe the result, and apply that learning to the next execution. More execution generates richer decision context; richer context improves future decisions; better decisions make execution more accurate and increasingly autonomous.

As foundation models commoditize, what companies learn through execution becomes a durable advantage. Customer-specific context improves performance and raises switching costs within an account, while recurring exceptions, execution patterns, and successful paths can improve performance across customers.

The moat isn’t the model. It’s what the company learns from repeatedly doing the work.

Expand Through Trust

Better outcomes build trust, and trust leads customers to hand over more work. What starts as one narrow outcome can expand into related work as each success gives customers confidence to delegate more. Unlike traditional software, which typically expands by selling another product or module, AI-native services can grow by owning more of the customer’s work.

Trust is what unlocks greater context and responsibility. Customers are unlikely to hand an AI every preference, relationship, system, and historical decision upfront. Instead, each successful outcome gives them confidence to share more context, connect more systems, and delegate more valuable work.

That context lowers the barrier to expansion. Preferences, relationships, systems, and prior decisions learned through one outcome are often relevant to the next. AI-native services therefore enter adjacent work with an existing understanding of how the customer operates rather than starting from zero.

This creates a fundamentally different expansion motion: customer pull rather than product push. Instead of acquiring a new customer or selling an unrelated module, AI-native services can grow by taking on more work for customers whose trust and context they have already earned. Each expansion increases revenue while adding context that makes the next expansion easier.

Evaluating AI-Native Services

Outcome ownership alone doesn’t make a venture-scale business. The key diligence question is whether that ownership creates a flywheel that can compound. Investors should pressure-test five characteristics:

1. Is this work customers would rather delegate than do themselves? 

The addressable market for AI-native services is constrained by what customers want to delegate, not what AI can execute. Willingness to delegate depends on two things: the value of handing off the process and the trust required to do so.

Customers value delegation most when the process is primarily a means to an end rather than part of the experience they want to preserve. AI could coordinate the purchase of a designer bag end-to-end, but many consumers value searching, comparing, and choosing for themselves. Technical capability alone does not make a process worth delegating.

Trust also determines what customers will hand over. As the consequences of getting an outcome wrong increase, customers need greater confidence before relinquishing control. This reinforces the wedge approach: start narrow, prove the outcome, and earn the right to own more.

At AV, we start by looking for coordination-heavy processes customers already want off their plate, particularly when the work requires reconciling fragmented context across people, systems, and sources. Examples include resolving IT issues, managing home repairs, or coordinating insurance claims.  If customers want to retain the process, there is no wedge to own and no flywheel to build.

2. Is there enough economic value in the outcome to support venture-scale revenue?

Existing service spend is the strongest signal because customers have already demonstrated both willingness to pay and allocated budget to getting the job done.

Not every delegable outcome supports meaningful spend. Consumers may happily delegate organizing old photos or maintaining a reading list, but completing those tasks carries little economic consequence. Outcomes tied to revenue, cost, risk, or essential responsibilities, such as legal work, home maintenance, or financial management, command much larger budgets.

As we invest, we look first for existing service spend. If customers do not already pay for the outcome, there should be a clear financial reason to start, such as generating revenue, reducing costs, or avoiding risk. The more customers already spend to get the job done, the more revenue an AI-native service can potentially capture.

3. Can the company own the outcome end-to-end? 

Outcome ownership requires more than removing work from a workflow. The product needs to close the loop from decision through execution to result. 

Risotto* makes the distinction concrete. Many IT support products optimize for ticket deflection, reducing work for IT without necessarily resolving the employee’s underlying issue. Risotto instead optimizes for auto-resolution, taking the request through to completion. Resolving the issue both delivers the outcome the employee needs and captures the decisions, actions, and exceptions required to achieve it.

Ownership must also become increasingly software-driven. Human intervention can scaffold execution early, as companies like Concorda* use experts alongside AI to deliver complex outcomes, but software should perform a growing share of the work over time. If human labor scales proportionally with outcomes delivered, the company may capture services revenue without achieving software economics.

For our investments, we track two trajectories: the share of the outcome the company owns should rise, while human effort per outcome should fall.

4. Does the workflow repeat frequently enough for learning to compound?

Frequency makes some AI-native service categories structurally better positioned to compound than others. Recurring workflows create more opportunities to execute, learn from the result, and apply that learning to the next outcome.

Leisure travel illustrates the distinction. Trip planning fits many characteristics of an attractive AI-native service: it is coordination-heavy, fragmented, and often desirable to delegate. Most consumers, however, travel only a handful of times per year, limiting how quickly the service can learn from repeated execution for each user. Hero Assistant* sits at the opposite extreme: by coordinating everyday workflows across calendars, tasks, reminders, and other daily tools, it has multiple opportunities to execute and learn each day.

We diligence the velocity of the learning loop, not usage frequency alone. The best-positioned workflows combine frequent execution with decisions that generate context that materially improves the next outcome.

5. Does that decision context transfer to valuable adjacent outcomes? 

A narrow wedge becomes venture-scale when the context it generates creates an advantage beyond the initial outcome. The most valuable context is reusable, giving the company a head start in adjacent work rather than requiring it to rebuild context and trust from zero.

LogCat* illustrates this progression. Parsing software logs gives it context on how systems behave and why failures occur. That context can support increasingly valuable outcomes, from identifying problems to recommending fixes and eventually executing them.

Not all decision context transfers. A company can become exceptionally good at one outcome while accumulating knowledge with little value beyond it. The venture opportunity depends on where that learning can travel.

For our portfolio companies, we  ask where the context becomes valuable next and to whom. Learning that transfers across customers strengthens the core product; learning that transfers across outcomes expands the market the company can address. The strongest wedges do both.

Final Takeaway

AI-native services create a rare venture dynamic: each outcome can generate revenue while making the next cheaper to deliver, better to execute, and easier to expand from. Growth doesn’t just make the company larger. It can improve the economics and defensibility of future growth.

For investors, the opportunity is to identify these flywheels early, when a company may still look like a narrow service but already has the ingredients to compound into a category-defining business.

1. Julien Bek, “Services: The New Software”, Sequoia Capital, March 5, 2026, https://sequoiacap.com/article/services-the-new-software.

2. Grand View Research, “U.S. Legal Services Market Size & Share | Report, 2030,” Grand View Research, accessed September 3, 2026, https://www.grandviewresearch.com/industry-analysis/us-legal-services-market-report.

3. Dataintelo, “Recruitment & Staffing Market Research Report, 2026–2034,” Dataintelo, published April 1, 2026, https://dataintelo.com/report/recruitment-staffing-market.

4. Market Report Analytics, “Home Services Market Analysis 2025 and Forecasts 2033: Unveiling Growth Opportunities,” Market Report Analytics, published August 16, 2025, https://www.marketreportanalytics.com/reports/home-services-market-4092.

*AV Portfolio Company. Portfolio companies referenced (Duckbill, Risotto, Concorda, Hero Assistant, LogCat) are shown for illustrative purposes only; they are not necessarily indicative of any AV fund or investor, and are not available to future investors, except potentially in the case of follow-on investments. Co-investors are shown for illustrative purposes only, do not reflect the universe of all organizations with which AV has co-invested, and do not necessarily represent future co-investors. The identity of a co-investor does not necessarily indicate investment quality or performance.

This communication is from Alumni Ventures, a for-profit venture capital company that is not affiliated with or endorsed by any school. It is not personalized advice, and AV only provides advice to its client funds. This communication is neither an offer to sell, nor a solicitation of an offer to purchase, any security. Such offers are made only pursuant to the formal offering documents for the fund(s) concerned, and describe significant risks and other material information that should be carefully considered before investing. For additional information, please see here. Achievement of investment objectives, including any amount of investment return, cannot be guaranteed. Co-investors are shown for illustrative purposes only, do not reflect all organizations with which AV co-invests, and do not necessarily indicate future co-investors. Example portfolio companies shown are not available to future investors, except potentially in the case of follow-on investments. Venture capital investing involves substantial risk, including risk of loss of all capital invested. Diversification cannot prevent investment loss; it is a strategy to mitigate investment risk. This communication includes forward-looking statements, generally consisting of any statement pertaining to any issue other than historical fact, including without limitation predictions, financial projections, the anticipated results of the execution of any plan or strategy, the expectation or belief of the speaker, or other events or circumstances to exist in the future. Forward-looking statements are not representations of actual fact, depend on certain assumptions that may not be realized, and are not guaranteed to occur. Any forward-looking statements included in this communication speak only as of the date of the communication. AV and its affiliates disclaim any obligation to update, amend, or alter such forward-looking statements, whether due to subsequent events, new information, or otherwise.