Sequoia Capital: The Next Trillion Companies Don't Sell Software, Sell Results Directly Author: Julien Bek Compiled: Deep Tide TechFlow Deep Tide Introduction: Julien Bek, a partner at Sequoia Capital, wrote a well structured article with the core argument that the next trillion dollar company will not sell software tools, but will directly sell work results. For every $1 spent on software, companies have to spend $6 on services. When AI makes the cost of "doing things" approach zero, the real opportunity is not in Copilot, but in Autopilot. He dismantled insurance, accounting, healthcare, and law one by one IT、 The automation opportunities in the procurement, recruitment, consulting and other service industries are accompanied by an opportunity matrix chart drawn from two dimensions: "intelligence vs judgment" and "outsourcing vs internal". It has reference value for both AI entrepreneurs and investors. The next trillion dollar company will be a software company disguised as a service company. Every founder of AI tools is asking the same question: What if the next version of Claude turns my product into a feature? This concern is not wrong. If you are selling tools, you are racing against models. But if you are selling the job itself, every improvement in the model makes your service faster, cheaper, and more difficult to compete with. A company may spend $10000 annually on QuickBooks and then spend $120000 to hire an accountant to settle the accounts. The next legendary company will directly settle your account for you. Intelligence vs. Judgment Writing Code is mainly about 'intelligence'. Knowing what to do next is a 'judgment'. Translating a requirements document into code, testing, and debugging: The rules are complex, but they are ultimately rules. Judgment is different. It requires experience and taste, as well as intuition accumulated through years of practice. Deciding what feature to do next, whether to owe technical debt, and when to release it before it's ready. A year ago, most cursor users used AI as autocomplete. Today, there are more tasks initiated by agents than by humans. Software engineering accounts for over half of all AI tool usage across all professions, while all other categories remain in the single digits. The reason is that software engineering is mainly an intellectual work. AI has crossed that line - it can autonomously complete most of its intellectual work, leaving judgment to humans. Software engineering arrived here first, but it will spread to every profession. Caption: The proportion of AI tool usage in various professions, software engineering far exceeds other categories. Copilot and Autopilot Copilot sell tools. Autopilot sells work. Until recently, AI models were still developing in terms of intelligence and judgment, so the correct path was to start with Copilot: putting AI in the hands of professionals and letting them decide how to use it. Harvey was sold to a law firm, and Rogo was sold to an investment bank. Professionals are customers, tools make them more efficient, and they are responsible for output. Today, models are already smart enough, and in some categories, the best starting point is to directly do Autopilot. Crosby sells to companies that need to draft NDAs, rather than to external legal advisors. Sell WithCoverage to CFOs in need of insurance, not to insurance brokers. What customers directly purchase is the result. In any profession, the work budget far exceeds the tool budget, and Autopilot can capture the work budget from day one. The higher the proportion of intelligence in a field, the faster Autopilot wins. Integrating today's judgments will become tomorrow's intelligence. As AI systems accumulate proprietary data in their respective fields to determine what they look like, the forefront will shift. Copilot and Autopilot will converge. The transition from Copilot to Autopilot has already begun in several categories. But the starting position is important because it determines where Autopilot can now win customers and begin accumulating data that ultimately allows it to handle judgment tasks as well. Autopilot strategy: Outsourcing is the entry point where for every $1 spent on software, $6 is spent on services. The TAM of Autopilot refers to all labor expenses within a category, including both internal and external costs. But the right starting point is where outsourcing already exists. If a task has been outsourced, it tells you three things. Firstly, the company has accepted that this task can be completed externally. Secondly, there is an existing budget subject that can be cleanly replaced. Thirdly, the buyer has already purchased the results. Replacing an outsourcing contract with an AI native service provider is equivalent to switching suppliers. Replacing internal employees is organizational restructuring. The strategy is to start with outsourced, intelligence intensive tasks. Get the distribution done. As AI accumulates data, it expands to internal, judgment intensive tasks. Outsourcing tasks are a wedge, while internal work is a long-term TAM. Crosby starts with NDA: a clearly defined task, primarily intellectual work, that most companies outsource to external lawyers. Ready made budget, clear scope, instant ROI, and frictionless replacement. By drawing the spectrum of "intelligence to judgment" and the ratio of "outsourcing to internal outsourcing" for each service vertical area on the opportunity map, a priority map can be obtained, with labor TAM in parentheses. The following list is not exhaustive. Caption: Autopilot opportunity matrix for various service verticals (distributed by intelligence/judgment ratio and outsourcing/in sourcing ratio) Insurance brokers ($14-200 billion). The largest market on this list. Standard commercial insurance is highly standardized: the added value of brokers is essentially a pure intellectual job of comparing prices and filling out forms between different insurers. The distribution layer is extremely fragmented, with thousands of small brokers running the same process, and no one controlling customer relationships. WithCoverage and Harper are interesting new entrants. Accounting and auditing (outsourcing only in the United States amounts to $50-80 billion). The United States has lost approximately 340000 accountants in the past five years, while demand has been increasing during the same period. 75% of CPAs are nearing retirement, with a long licensing path and starting salaries lagging behind those in the technology and finance industries. This structural shortage is driving accounting firms to adopt AI faster than almost all other professions. Rillet is building an AI native ERP for direct billing. Basis started with Copilot in accounting. Medical revenue cycle management (outsourced portion of $50-80 billion in the United States). When you hear about 'healthcare', you may feel that judgment is intensive, but the accounting layer is almost purely intellectual work. Medical coding is the process of translating clinical notes into approximately 70000 standardized ICD-10 codes. The rules are complex, but they are ultimately rules. Outsourcing is already mature and billed based on results. Autopilot only needs to do the same thing at a lower cost. The former goes the farthest. Claims and loss assessment (including TPA of $50-80 billion). On the other side of the insurance policy, claims assessment is another independent Autopilot scenario. The claims for standard insurance are determined based on the policy language and damage list, and reserves are set using actuarial tables. The loss assessment team is aging and there is no one to make up for it. The market is heavily outsourced to independent loss adjusters and TPAs such as Crawford and Sedgwick. There are at least two different Autopilot opportunities in an industry. Pace is doing Autopilot for claims processing, while Strala is doing AI native TPA. Tax consulting ($30-35 billion). The CPA license system creates a regulatory moat, but 80% -90% of the work at the bottom is intellectual work. For every additional jurisdiction covered by Tax Autopilot, the data moat becomes even deeper. The complexity of multiple jurisdictions is precisely the reason why small and medium-sized enterprises outsource, as no internal accountant can fully cover it. TaxGPT is an early entrant, with Skalar and Ravical in Europe. Legal transactional work (20-25 billion US dollars). Contract drafting NDA、 Regulatory declaration: high intelligence proportion, conventional outsourcing. The work output is sufficiently standardized and the quality is verifiable, so buyers can trust AI output without requiring deep legal expertise. Harvey is a rising leader rapidly shifting towards Autopilot; Crosby and Lawhive are new entrants to Autopilot native. IT hosting services (over $100 billion). Every small and medium-sized enterprise outsources IT. Patching, monitoring, user configuration, and alarm diversion: intellectual work runs repeatedly in thousands of identical environments. Existing software layers (ConnectWise, Datto) sell tools to MSP. No one has directly sold 'your IT is running' as a result to the company yet. Edra is automating IT processes, while Serval is automating IT support. Supply chain and procurement (over $200 billion). Most companies only negotiate seriously with the top 20% of suppliers. Long tail suppliers are completely neglected because it's not cost-effective to have people do this. Contract leakage accounts for 2% -5% of total procurement expenses. The entry point is the abandoned job: there are no budget subjects to argue for, no incumbents to replace, only money picked up in vain. Magentic is doing direct procurement AI, while AskLio is doing indirect procurement. Tacto is simultaneously building a recording system and Copilot for the mid size market. Recruitment and manpower dispatch (over 200 billion US dollars). The largest service market on this list. The top of the recruitment funnel (screening, matching, reaching) is purely intellectual work, but the selection and evaluation of cultural fit are judgments accumulated through years of pattern recognition. The entry point of Autopilot is in high volume, low judgment positions, where the matching is standardized. Juicebox, Mercor, Jack&Jill are emerging leaders building the entire spectrum. Management consulting ($300-400 billion). A huge market, but the job is mainly judgment. An interesting question is whether AI can break down consulting into intelligence components (data collection, benchmarking analysis) and judgment components (strategic recommendations), with the intelligence layer automated and the judgment layer left to humans. The best candidate is yet to be determined. The fastest growing AI company in 2025 is Copilot. In 2026, many will attempt to switch to Autopilot. They have product and customer awareness. But they also face the dilemma of innovators: selling jobs means kicking their customers out of the job. This is the opportunity window for pure Autopilot company.