服务：新形态的软件

ByJulien BekPublished March5,2026 The next $1T company will be a software company masquerading as a servicesfirm. Every founder building an AI tool is asking the same question: what happens when thenext version of Claude makes my product a feature? They’re right to worry. If you sellthe tool, you’re in a race against the model. But if you sell the work, every Intelligence vs Judgement Translating a spec into code, testing, debugging: the rules are complex but they arerules. Judgement is different. It requires experience and taste, instinct built on years ofpractice. Deciding which feature to build next, whether to take on tech debt, when to A year ago, most Cursor users treated AI as autocomplete. Today, more tasks arestarted by agents than by humans. Software engineering accounts for over half of all AItool usage across professions. Every other category is still in single digits. The reason is Copilots and Autopilots A copilot sells the tool. An autopilot sells the work. Until recently, AI models were still developing intelligence and judgement, so the rightapproach was to build a copilotfirst: put AI in the hands of a professional and let themdecide what to do with it. Harvey sells to lawfirms. Rogo sells to investment banks. Today, the models are intelligent enough that in some categories the best place tostart is as an autopilot. Crosby sells to the company that needs an NDA drafted, not tooutside counsel. WithCoverage sells to the CFO who needs insurance, not to the The higher the intelligence ratio in anyfield, the sooner autopilots will win. The Convergence Today’s judgement will become tomorrow’s intelligence. As AI systems accumulateproprietary data about what good judgement looks like in their domain, the frontierwill shift. Copilots and autopilots will converge. The copilot-to-autopilot transition has The Autopilot Playbook: Outsourcing as the Wedge For every dollar spent on software, six are spent on services. The total addressable market for autopilots is all labour spend in a category, insourcedand outsourced combined. But the right place to start is where outsourcing already If a task is already outsourced, it tells you three things. One, the company hasaccepted that this work can be done externally. Two, there’s an existing budget linethat can be substituted cleanly. Three, the buyer is already purchasing an outcome. The playbook: companies should start with the outsourced, intelligence-heavy task.Nail distribution. Expand toward the insourced, judgement-heavy work as the AI Crosby started with NDAs: a well-defined task, primarily intelligence, that mostcompanies already outsource to external counsel. The budget exists, the scope is clear, Opportunity Map Plotting every services vertical on an intelligence-to-judgement spectrum andoutsourced-to-insourced ratio produces a priority map with labour TAM in brackets. Insurance brokerage ($140-200B).The largest dollar market on this list. Standardcommercial lines are highly standardised: the broker’s value-add is essentiallyshopping across carriers andfilling forms, pure intelligence work. The distribution Accounting and audit ($50-80B outsourced in the US alone).The US has lostroughly340,000accountants overfive years while demand has grown.75% of CPAs arenearing retirement, the licensing path is long, and starting salaries lag tech and Healthcare revenue cycle ($50-80B outsourced in US). People hear“healthcare”and assume it’s judgement-heavy, but the billing layer is almost pure intelligence.Medical coding is translating clinical notes into ~70,000standardised ICD-10codes. outcome-based. An autopilot just has to do the same thing at lower cost. Anterior is Claims adjusting ($50-80B including TPAs).On the other side of the insurancepolicy, claims adjusting is a separate autopilot surface. Standard-line claims aresettled by interpreting policy language against damage schedules and setting reservesusing actuarial tables. The adjuster workforce is aging out and nobody’s replacing Tax advisory ($30-35B).CPA licensing creates a regulatory moat, but80-90% of theunderlying work is intelligence. Every additional jurisdiction a tax autopilot handlesdeepens its data moat. Multi-jurisdiction complexity is exactly what SMBs outsource Legal, transactional work ($20-25B).Contract drafting, NDAs, regulatoryfilings: highintelligence, routinely outsourced. The work product is standardised enough thatquality is verifiable, so the buyer can trust AI output without deep legal expertise. IT managed services ($100B+).Every SMB outsources its IT. Patching, monitoring,user provisioning, alert triage: intelligence work running on repeat across thousandsof identical environments. The existing software layer (ConnectWise, Datto) sells tools Supply chain and procurement ($200B+).Most enterprises negotiate seriously withonly their top20% of suppliers. The long tail gets zero attention because it’s noteconomical to have humans do the work. Contract