The core premise of the SaaS era is collapsing. For two decades, venture capital operated on a simple truth. Writing proprietary code creates a defensible, high margin economic moat. Today, that moat is dry.
As the marginal cost of software development drops toward zero, traditional SaaS vendors find themselves in an unwinnable race against foundational models. The next venture capital thesis is not about building marginally better software to assist human workers. It is about deploying autonomous systems that replace the software and the labor simultaneously.
Sequoia Capital recently noted that the next EUR 1 trillion company will be a software company masquerading as a services firm. We are entering the era of service as a software.
If you sell a tool, you are in a race against the model. If you sell the work, every improvement in the model makes your service faster, cheaper, and harder to compete with. In the AI native economy, the strategic imperative is no longer selling the software. It is selling the completed outcome.
Why Now? The Shift from Copilots to Autopilots
The most critical question is why this has not been built before now.
Until recently, AI models were still developing basic capabilities. The right approach was to build a copilot. You put AI in the hands of a professional and let them decide what to do with it. The professional remained the customer, and they took responsibility for the output.
Today, models have crossed the threshold where they can handle the intelligence work autonomously, leaving only the complex judgement to humans. This allows founders to build autopilots that sell the work directly.
The Math Behind the Moat
Strong investments are built on bottom up realities. To understand the scale of margin expansion available, we must calculate the exact difference between software spend and services spend.
Consider European accounting and tax advisory. Europe is home to roughly 25 million small and medium businesses.1 If they spend an average of appr. EUR 1,000 annually on outsourced bookkeeping and tax advisory, it creates a EUR 25 billion services market. Meanwhile, the entire European SMB accounting software market is capped at roughly EUR 6 billion.1 By absorbing the service layer rather than just selling the software layer, an AI native entrant expands the Total Addressable Market by a factor of over four.
The multiple is even more extreme in the legal sector. The European legal services market is projected to reach EUR 265 billion in 2026.2 In contrast, the European legal technology software market sits at roughly EUR 6 billion. Capturing a fraction of this highly fragmented labor spend completely dwarfs the upside of selling a traditional B2B SaaS tool, as the services market is over 44 times larger than the software market.
The 3H Framework: Assessing the Automation Asymptote
To build a category defining company in this space, founders need a rigorous mental model to evaluate which service verticals are ripe for disruption. To assess a vertical's margin potential, we use the 3H Framework to evaluate the enterprise workflow against three core constraints.
- Hands (The Physical Constraint): Does the task require bespoke, unstructured physical interaction in the real world?
- Hearts (The Judgment Constraint): Does the outcome rely on deep emotional intelligence, empathy, complex negotiation, or the establishment of human trust?
- Handcuffs (The Liability Constraint): Does the final deliverable require a human to assume regulatory, legal, or ethical liability?
The answers to these questions define the boundary between machine execution and human necessity. This boundary is the Automation Asymptote. Let us apply the 3H framework to four distinct verticals to demonstrate how to map agentic capability to margin potential.
Real Estate Brokerage
In real estate brokerage, the intelligence layer is perfectly suited for autonomous agents. AI handles lead generation, comparative market analysis, marketing distributions, and document preparation. Humans are only strictly required for physical viewings, navigating the nuanced psychology of the final negotiation, and managing the notary sign off. Because the machine executes the vast majority of the workflow, the automation asymptote reaches 70 to 80 percent. The human retains the remaining 20 to 30 percent as a physical presence and trust premium.
Insurance Brokerage
When applied to insurance brokerage and claims administration, the margin profile shifts dramatically. Workflows here are heavily governed by strict contractual parameters, probabilistic models, and structured documentation. An AI agent can handle the entire lifecycle autonomously. Zero Hands and zero Hearts are required. Consequently, the maximum agentic capability approaches 100 percent only limited by some handcuffs regarding broker liability. This allows the platform to drive almost pure software margins in a legacy service industry.
Audit and Assurance
Audit represents an extreme Handcuffs constraint. Specialized AI agents can ingest massive financial ledgers and turn weeks of manual fieldwork into days of intelligent analysis. But the automation asymptote is hit exactly where professional skepticism begins. A human is legally and ethically required to determine if an anomaly constitutes fraud and to formally sign the audit report. Margin potential is strictly dictated by this split between high machine intelligence and absolute human liability. This regulatory requirement creates an agentic ceiling, capping the agentic capability around 60 to 70 percent.
Tax Advisory
Tax advisory shares a similar profile to audit but with distinct workflow advantages. Agentic AI can autonomously ingest financial records, automate compliance checks, extract data from receipts, and classify complex documents. However, the human tax professional must remain in the loop for strategic advisory, complex cross border structuring, and assuming regulatory liability before tax authorities. This caps the agentic capability around 75 to 85 percent, allowing a platform to capture massive efficiency gains while retaining a human expert for the final legal sign off.
Three Vectors of Market Entry
Understanding a vertical's margin cap is only the diagnostic first step. The critical challenge is the go to market wedge. We see three distinct pathways to architecting an AI native service firm.
1. The Business in a Box
Effectively a Shopify for Services, you provide independent professionals with a toolkit to run their entire business. As you deploy increasingly autonomous agents, your platform's take rate increases toward the vertical's maximum agentic capability. Because of massive productivity gains, the humans earn more absolute dollars. This masks the aggressive margin extraction.
2. The Full Stack Autonomous Service Provider
You bypass the software vendor model entirely. You hire domain experts and incentivize them to systematically agentify their own workflows. Because you own the end to end service delivery, every technological breakthrough drops directly to your bottom line.
3. The AI Enabled Roll Up
This represents the ultimate convergence of venture capital and private equity. You acquire legacy service companies operating at 5 to 15 percent margins and systematically replace their back office operations with agentic AI. You generate compounding returns by aggressively automating administrative work, radically expanding the gross margins of the acquired firms and turning them into highly scalable tech assets. Scaling this requires a dual threat leadership structure: marrying aggressive M&A operational expertise with elite AI product engineering. Europe presents a unique environment for this transition. The regional economy relies heavily on fragmented, analog businesses that are currently navigating a massive generational handover. These traditional firms are ripe for technological consolidation. Moreover, agentic systems naturally dissolve Europe's historical borders by instantly adapting to local languages and compliance rules, allowing a domestic service provider to scale across the continent without the usual friction of hiring local, multilingual staff.
Measuring the Autonomous Enterprise
Before true margin expansion materializes, the underlying unit economics must be actively managed. The defining metrics of the AI native firm are no longer user engagement or seat count, but the sheer volume of digital work hours produced and the percentage of workflows resolved entirely autonomously.`
The Monetization Imperative: Outcome Based Pricing
Executing these models requires abandoning the foundational metric of the SaaS era. You must kill the seat license.
When software acts as an autonomous agent replacing human labor, charging per human seat is self defeating. To survive and scale, founders must adopt Outcome Based Pricing. You charge for the completed work, the resolved claim, or the finalized tax return. This aligns vendor revenue directly with the customer value created. You transform AI from a fixed IT cost into a variable cost of operations. When pricing is tied to the actual work delivered, vendors capture a true percentage of the economic value generated by the agent, rather than fighting over a shrinking software budget.`
The Pre Mortem: What Breaks This Thesis?
A rigorous pre mortem is required to outline what could kill the strategy. What are the existential risks that could disrupt the transition to the AI native economy?`
1. Total Addressable Market Deflation
The most severe risk is value destruction rather than value capture. If AI drops the cost of delivering a service by 90 percent, the overall market size might shrink dramatically. A EUR 12 trillion global service market could deflate into a EUR 2 trillion market. If prices collapse faster than volume increases, AI native firms will fight over a much smaller revenue pool.`
2. Incumbent Data Moats and API Lockouts
Agentic AI requires deep access to historical data and read write access to enterprise systems. Legacy systems of record could restrict API access or aggressively price API calls to starve AI native startups of the required data. If incumbents successfully block access, startups will be unable to execute the full service loop. Computer use agents could act as a workaround to operate software interfaces directly, but this approach is currently cumbersome and fragile.
3. The Regulatory Handcuffs Ceiling
In heavily regulated industries, ethical concerns, data privacy issues, and the need for human liability remain severe barriers to adoption. If regulators mandate strict human in the loop requirements for critical services, the automation asymptote will artificially stall. This would cap gross margins and destroy the software economics that make these businesses venture backable.
Architecting the Future at The Delta
The transition from software to services is inevitable. At The Delta, we are not just observing this transition. We are building the infrastructure for it.
While incumbents are constrained by the biological friction of hiring, training, and managing humans, an AI native firm simply clones its best digital performer and provisions more compute. Our venture studio model is designed to be the premier launchpad for this exact transition, providing the capital, engineering talent, and operational frameworks necessary to build full stack autonomous businesses.`
If you are a founder ready to abandon legacy SaaS models and architect the autonomous future, get in touch with our venture studio. It is time to build the future of service businesses. Furthermore, very soon robotics will take over much of the Hands limitation as well.
Written by Julian Teicke
Chairman
