The services-as-software thesis argues that AI is inverting the dominant technology business model of the past twenty years. Instead of selling software tools that help customers perform work (software-as-a-service), companies will use AI to perform the work internally and sell the finished output (services-as-software). The customer buys the outcome, not the tool.

Foundation Capital formalized this thesis in a 2025 research brief, estimating the total addressable market at $4.6 trillion: the combined global spend on human-delivered professional services that AI can now perform or substantially automate. Their analysis found that 60-80% of agency and professional services revenue comes from execution, not strategy. AI collapses the cost of execution while preserving the value of the strategic layer.

Y Combinator validated the thesis independently. In its Spring 2026 Request for Startups, YC listed "AI-Native Agencies" as a priority investment category. Group Partner Aaron Epstein wrote the brief, arguing that agencies have historically repelled venture capital because growth is tied to headcount. AI breaks the constraint. An AI-native agency can achieve software-like margins (60-80%) while delivering services outcomes.

What is the difference between SaaS and services-as-software?

The distinction is not branding. It is architectural. It determines who does the work, how the business scales, and what the customer pays for.

The inversion is important for understanding what is happening in the agency market. SaaS companies have spent two decades selling productivity tools to agency workers. Services-as-software companies are replacing the agency workers with AI agents and selling the finished work directly to clients.

Where does the services-as-software thesis come from?

The intellectual lineage runs through three sources, each building on the previous one.

Foundation Capital (2025): Ashu Garg published the foundational analysis. The core argument: every professional service that follows a repeatable process is vulnerable to software automation. Accounting, legal review, customer support, marketing execution, media relations. The $4.6 trillion figure represents the aggregate spend on these services globally. Foundation Capital began investing in companies executing the thesis before publishing the research.

Y Combinator (Spring 2026): Aaron Epstein's RFS brief applied the thesis to agencies specifically. His framing: "The next big companies may not sell software. They'll do the work." Epstein argued that agencies are the clearest application of the thesis because agency work is process-heavy, execution-dense, and scales poorly with humans. AI-native agencies fix all three constraints. YC began accepting AI-native agency applications for its Spring 2026 batch.

UBS / Accenture (2026): UBS analyst Kevin McVeigh projected that the convergence of AI, software, and services into outcome-based integrated systems could create a $1.5 trillion "services-as-software" opportunity by 2035. Accenture restructured its entire organization under a unified "Reinvention Services" model to position for this shift. When a $64 billion consulting firm reorganizes around a thesis, the thesis has institutional weight.

Which companies are executing the services-as-software thesis?

Funded companies are proving the model across verticals. Each applies the same structural logic: use AI to perform the execution work, sell the completed output to clients, scale without proportional headcount growth.

• 14.ai (YC W26, $3M seed): AI-native customer support agency. Replaces ticketing systems, AI software add-ons, and human support agents with a single contract. Backed by General Catalyst, SV Angel, and founders of Dropbox and Slack. (TechCrunch, March 2026)

• Mega ($11.5M Series A, a16z): AI-native marketing agency for SMBs. Zero to $10M revenue in ten months. Positions as "an enterprise-grade growth team, without the agency."

• Multiply ($9.5M, Mayfield): AI agents for B2B advertising. Emerged from stealth March 2026.

• Shadow (shadow.inc): AI-native communications infrastructure. AI agents perform research, media targeting, content production, and competitive intelligence for PR agencies and technology companies. Clients review finished work in under two hours per month.

• EPAM Empathy Lab (enterprise): AI-native agency for brand growth, launched in North America February 2026. Part of EPAM Systems ($3.4B annual revenue), signaling that the model scales beyond startups.

What does services-as-software mean for communications?

The communications industry fits the services-as-software pattern precisely. The work is process-intensive: research, media list building, pitch drafting, coverage monitoring, award applications, content production. Most of this work follows repeatable patterns. Senior strategists add the most value in judgment, relationships, and narrative framing, but execution consumes 60-80% of billable hours.

The traditional PR agency model has structural constraints that the thesis directly addresses:

• Revenue scales with headcount. Adding a new client requires hiring or reallocating an account team. AI-native communications infrastructure breaks this constraint by having AI agents handle execution.

• Margins are thin. Agency gross margins typically range from 30-50%. AI-native models push margins toward 60-80% by replacing execution labor with compute.

• Quality is inconsistent. Work quality depends on which humans are assigned to the account. AI-native models deliver consistent quality because the same agent architecture serves every client.

The thesis does not argue that AI replaces the strategist. It argues that AI replaces the execution layer, freeing the strategist to focus on judgment and relationships. The agencies that adopt this model can serve more clients at higher quality with fewer people. Those that do not will compete against AI-native entrants operating at fundamentally different economics. For a full definition of the model, see What Is an AI-Native Agency? For a practical roadmap to making the transition, see How a PR Agency Becomes AI-Native.

Frequently asked questions

Is services-as-software the same as AI-as-a-service?

No. AI-as-a-service (AIaaS) sells access to AI models or infrastructure (OpenAI's API, AWS Bedrock, Google Vertex AI). The customer builds their own applications on top. Services-as-software uses AI internally and sells the completed work product. The customer never touches the AI. Foundation Capital's thesis is specifically about this distinction: the value shifts from selling the tool to selling the outcome.

How big is the services-as-software market?

Foundation Capital estimates the total addressable market at $4.6 trillion, representing global spend on human-delivered professional services. UBS projects the segment could reach $1.5 trillion by 2035 as Accenture and similar firms restructure around the model. Both estimates focus on services where execution follows repeatable processes: accounting, legal review, customer support, marketing, media relations, and consulting.

Why are VCs funding agencies now when they avoided them before?

Traditional agencies scale linearly with headcount, producing 30-50% gross margins. Venture capital requires non-linear scaling. AI-native agencies break the headcount constraint: revenue can grow 10x without 10x the employees because AI agents handle execution. Y Combinator's Aaron Epstein and Foundation Capital's Ashu Garg both argue this changes the margin structure enough to make agencies venture-backable for the first time.

What is the difference between Foundation Capital's thesis and Y Combinator's?

Foundation Capital frames the thesis as a macroeconomic shift: $4.6 trillion in services spend moving from human delivery to software delivery. Y Combinator frames it as an agency-specific opportunity: AI-native agencies are the vehicle that captures the shift. The theses are complementary. Foundation Capital identifies the market. Y Combinator identifies the business model.

About the author

Jessen Gibbs is co-founder and CEO of Shadow, an autonomous communications infrastructure company. Before founding Shadow, Jessen [relevant background placeholder]. Shadow's AI agents handle communications work for PR agencies and technology companies.

Last updated: March 2026. Published by Shadow (shadow.inc). Shadow is an AI-native communications infrastructure company. Market estimates sourced from Foundation Capital and UBS as cited. Funding data sourced from TechCrunch, company announcements, and Crunchbase. Pricing and projections reflect published figures as of March 2026 and may change.