Sequoia partner Julien Bek argues the next $1T company will be a software company masquerading as a services firm. If you sell the work instead of the tool, every model improvement makes you faster and cheaper — not obsolete.

Services: The New Software

Julien Bek, Partner at Sequoia Capital — March 2026 | 2M views, 3.6K likes, 662 reposts

The Core Thesis

"The next $1T company will be a software company masquerading as a services firm."

If you sell the tool, you're in a race against the model. But if you sell the work, every improvement in the model makes your service faster, cheaper, and harder to compete with.

A company might spend $10K/year for QuickBooks and $120K on an accountant to close the books. The next legendary company will just close the books.

Intelligence vs Judgement

This is the key distinction:

Intelligence — writing code, translating specs, testing, debugging. Rules are complex but they are rules. AI handles this now.
Judgement — knowing what to build next, whether to take tech debt, when to ship. Built on years of experience and taste.

AI has crossed the intelligence threshold for software engineering. It is coming to every profession.

"Instead of 10 humans per AI we will have 10 AI per human."

Copilots vs Autopilots

| | Copilot | Autopilot | |---|---|---| | Sells | The tool | The work | | Customer | The professional | The business buying the outcome | | Budget captured | Tool budget | Work budget (6x larger) | | Examples | Harvey (law firms), Rogo (banks) | Crosby (NDA drafting), WithCoverage (insurance) |

The playbook: Start with the outsourced, intelligence-heavy task. Nail distribution. Expand toward the insourced, judgement-heavy work as the AI compounds.

"Replacing an outsourcing contract with an AI-native services provider is a vendor swap. Replacing headcount is a reorg."

The Opportunity Map

Bek maps every services vertical on an intelligence-to-judgement spectrum × outsourced-to-insourced ratio:

Healthcare revenue cycle ($50-80B) — medical coding is translating clinical notes into ICD-10 codes. The rules are complex but they are rules. Already outsourced and outcome-based.
Insurance brokerage ($140-200B) — standard commercial lines are essentially shopping across carriers and filling forms.
Accounting and audit ($50-80B) — structural shortage of CPAs is pushing firms to accept AI faster than almost any other profession.
IT managed services ($100B+) — patching, monitoring, user provisioning across thousands of identical environments.
Legal, transactional ($20-25B) — contract drafting, NDAs, regulatory filings.

"For every dollar spent on software, six are spent on services."

The Convergence

Today's judgement becomes tomorrow's intelligence. As AI systems accumulate proprietary data about what good judgement looks like in their domain, the frontier shifts. The moat is compounding data + domain expertise.

The Sequoia autopilot map is actively being assembled (see Julien's follow-up: "Turns out there were a lot more great founders building autopilots than I thought.")

Why This Matters Here

Healthcare is explicitly named as a prime autopilot target — specifically revenue cycle and claims. But the same logic applies higher up the stack: provider intelligence, BD data workflows, market analysis. The outsourcing already exists (consulting firms, data vendors, analyst teams). The work budget dwarfs the tool budget.

The wedge is the outsourced, intelligence-heavy task. The long-term TAM is everything else.