Last week I spent the afternoon at Emergence Capital’s AI-Native Services (AINS) summit in San Francisco. They handed everyone a physical playbook on the way in—”The AI-Native Services Playbook”—with a tagline printed on the cover: sell the fish, not the pole.
It stuck with me. That one line captures the single most important strategic choice every AI company is facing right now. And as a result, it’s influencing venture strategy across stages.
If you sell the AI tool, you’re in a race against the model. Every new Claude or GPT release is a threat. If you sell the outcome—the fish—every model improvement makes your service faster, cheaper, and harder to compete with. Better models flow to you, not against you.
This is the AINS (and yes, pulling from my inner Mean Girls, we’re in support of making “AINS” happen) thesis in one sentence. Emergence (here) and Sequoia (here) have both written compellingly about why this model is having a moment—and Thrive recently published a piece called “Long Humans” that approaches the same question from a different angle. I’d encourage you to read all three. Rather than recap their frameworks, I want to share Pear’s lens: where we’re placing bets, why certain verticals are structurally better suited than others, and what we’re actively looking for.
The short version of the framework
Here’s what AINS is and isn’t according to Emergence and Sequoia.
For every dollar spent on software in professional services, six are spent on human labor. Legal ($396B services vs. $10B software), wealth advisory ($260B vs. $7B), accounting ($155B vs. $6B), recruiting ($190B vs. $5B). The software layer is comically thin.
Two things follow from this: first, you’re displacing an existing budget, not creating one—you only have to win the “choose us” decision, not the “spend on this” decision. Second, gross margins of 70%+ are achievable when AI does the heavy lifting. If your margins aren’t expanding as you scale, you haven’t built an AINS company—you’ve built a better-staffed services firm with venture capital on the balance sheet. Emergence calls the latter “Mirage PMF.” And, it’s the most common trap founders fall into.
One important nuance the Thrive piece captures well: AINS isn’t really about replacing humans. It’s about raising the ceiling on what a trusted practitioner can do. Thrive uses the electric motor analogy—when factories first electrified, they just swapped out the steam engine for one big motor and got marginal gains. The real leap came when they redesigned the factory floor around smaller, distributed motors.
AI is a similar moment. The biggest gains won’t come from bolting AI onto existing workflows. They’ll come from redesigning the work entirely—and that’s what the best AINS founders are doing from scratch. And importantly, many of the cogs in that workstream, the “doers,” are the service practitioners themselves. I.e. AI-native law firms employing legal teams in house, leveraging AI tooling for efficiency and accuracy.
Pear’s bets and what we’re pattern-matching on
We’ve been deploying against this thesis with a specific founder archetype in mind: someone who has seen the service from the inside—enough domain credibility to earn trust from buyers and recruit senior practitioners—but who is fundamentally a product builder, not a practitioner. Not nostalgic about the old model. Ready to reimagine it from first principles. And ideally, they’re paired with a top CTO who can attract and train elite technical talent.
That combination is rare. In traditional SaaS, deep domain expertise can actually be a liability—it anchors founders to incumbent workflows rather than reimagining them. In AINS, domain credibility is a sales and hiring asset—buyers trust you faster, ex-practitioners know exactly which tasks to automate first—but only if paired with genuine technical skills and ambition.
Here are a few recent investments we’ve made at Pear against this thesis:
Andera: AI-native SOX audit (Finance)
SOX compliance is one of the most labor-intensive, repetitive, and expensive obligations for any public company—and the work is almost entirely rules-based. Control testing, evidence gathering, workpaper documentation: audit teams spend thousands of hours cross-checking spreadsheets and chasing down control owners via email, while Big 4 and mid-tier firms charge hundreds of thousands to do the same.
Andera automates the full SOX testing workflow, cutting costs by 70% and freeing audit teams to focus on judgment-level work rather than data wrangling. The budget is mandated by law, the outcome is unambiguous (controls tested, workpapers filed), and as AI handles more of the testing layer, margins expand naturally. Aryo and the team just announced a $37M Series A led by Lightspeed with participation from Bain Capital and Pear – a timely validation that the market is ready for this.
Veros: AI-native TrustCo (Finance)
Estate planning is one of the highest-trust, most relationship-intensive professional services categories—and one of the least disrupted. A basic revocable trust runs $2,000–$10,000 at a traditional firm and takes weeks (including dozens of emails, faxes, notarized documents, in person appointments, etc.).
More importantly, the vast majority of Americans who should have a trust, don’t. That’s not just a cost problem; it’s an access problem. The intelligence layer—drafting documents, structuring around an asset inventory, understanding family dynamics—is well-suited for AI. The judgment layer (what structure is right for you) stays human, but is dramatically improved when informed by AI synthesis across thousands of similar situations.
Veros is building the system of record for people’s legal and financial lives, starting with trusts and expanding from there. The Veros team fits the mold perfectly: Composed of a Yale grad and Stanford GSBer, Stanford CS grad, and a licensed South Dakota attorney and experienced senior trust officer.
Quinn: AI-native commercial insurance brokerage
Commercial insurance is a $140–200B market. The broker’s job—understand the risk, shop across carriers, advise on coverage, fill forms—is highly standardized intelligence work. The distribution layer is incredibly fragmented: tens of thousands of independent brokers, no single incumbent controlling the customer relationship. That fragmentation is the wedge.
Quinn is rebuilding the brokerage from scratch, capturing the work budget directly rather than selling tools to brokers. The Quinn co-founders are both technical and each grew up in the world of commercial insurance.
We also have early bets in AI-native hardware procurement, AI-native financial audit and AI-native banking ops—all in stealth, all in verticals with enormous services spend and very thin software penetration.
Where we’re still looking
Four verticals stand out as our highest priorities.
Legal
Legal has the highest services-to-software ratio of any major vertical (40x). And despite feeling bespoke, 80% of the work at the early-stage company layer is genuinely routine: formation documents, equity grants, commercial contracts, board consents. It’s intelligence work masquerading as custom work. The regulatory moat is real – bar licensure and liability mean model labs won’t attack this directly. The founder archetype is well-defined: someone with BigLaw pedigree (Cooley, Wilson, Latham) who can recruit practitioners and earn buyer trust immediately, but who left because they saw the industry was ripe to be rebuilt. We’re actively evaluating companies here.
Finance
Accounting and financial services represent $155B in services spend against just $6B in software —a 26x ratio—and the structural conditions for disruption are arguably more acute here than in any other vertical. The accounting profession is in a slow-motion workforce crisis: the U.S. has lost roughly 340,000 CPAs over the past five years, 75% of credentialed accountants are nearing retirement, and the pipeline of new entrants isn’t close to replacing them. Firms are being forced to accept AI faster than almost any other profession simply because they have no other choice.
The work itself—bookkeeping, tax prep, financial close, audit—is highly rules-based and already heavily outsourced by SMBs and growing companies, meaning the procurement motion is familiar. Outcomes are unambiguously measurable: did the books close? Did the audit pass? Was the return filed accurately? We have an early stealth bet in AI-native financial audit, but we’re looking for additional exposure across the broader finance stack—particularly in SMB accounting and tax advisory, where the incumbent service is fragmented, the buyer is cost-sensitive, and the value of speed and accuracy is immediately quantifiable.
Healthcare
Healthcare is enormous and often overlooked in AINS conversations, partly because “healthcare” sounds judgment-heavy. But the billing and revenue cycle layer is almost pure intelligence. Medical coding means translating clinical notes into roughly 70,000 standardized ICD-10 codes— the rules are extraordinarily complex, but they are rules. Revenue cycle management is already heavily outsourced by hospitals and health systems, meaning the procurement motion is familiar. There’s also a structural workforce dynamic: the healthcare admin workforce is aging out faster than it’s being replaced, creating urgency on the buyer side that accelerates adoption. And regulation (HIPAA, CMS) paradoxically helps founders—it keeps foundation model providers from competing directly.
Recruiting & Staffing
Recruiting is one of the most obvious categories to target with the AINS approach. It has $190B in services spend, $5B in software—a 38x ratio nearly matching legal. The outcome is highly measurable (Did the hire work out? Are they still there at 90 days?). The budget is already fully allocated—no line item creation required. The top of the funnel (screening, matching, outreach) is essentially pure intelligence work, making it a natural AI wedge. The long-term expansion is toward harder-to-fill, judgment-heavy placements as the data flywheel compounds.
Why now
The standard “why now” is model capability. AI can finally do the work, not just assist with it. That’s true, but there’s a more specific dynamic we’re watching.
The buyers we care about—startup founders, SMBs, growing companies—are making services vendor decisions right now, before category defaults are established. In enterprise SaaS, the Salesforces and Workdays won their markets over decades. In AINS, the category leaders for legal, insurance, accounting, and recruiting services for the next generation of companies are being chosen in the next few years. First-mover advantages in trust and distribution compound fast in services in a way they don’t in software.
There’s also a founder availability dynamic. The people best suited to build these companies—practitioners with 1-4 within established services firms who came up in the AI era—are starting to leave. They have the domain credibility. They’ve watched AI transform adjacent industries from the inside. And they’re increasingly willing to bet that the old model is worth tearing down. That wave is just beginning.
If you’re building an AI-native services company, or know someone who is, especially in legal, finance, healthcare, or recruiting, I’d love to talk. ryan@pear.vc