Sell the Work, Not the Seat: Vertical SaaS Just Became Vertical Labor
The 2026 cohort isn't building dashboards for body shops and metal finishers — it's answering their phones, fighting their insurers, and taking the services budget instead of the software line item.
By PlatoSeed Research · grounded in the live corpus
The wave right now
For a decade, "vertical SaaS" meant building the system of record for an industry and renting it out per seat. The 2023–2026 wave is doing something different: it sells the work itself. The new vertical companies don't give the body shop a dashboard — they answer its phones, fight its insurance denials, and key orders into the ERP it already hates. Rote turns insurer denials into supplement arguments with P-page citations in ten minutes; Walter reads documents and writes directly into manufacturing ERPs; Moritz isn't legal software at all — it's a law firm where AI does 80% of the work.
This changes who can be a customer. Metal finishers, skip-hire fleets, and collision shops never bought software well, because software meant a second job operating it. Labor that operates itself is a different sale — and it's why this cohort is crawling into industries the last SaaS wave wrote off as too small or too analog.
The landscape today
1. Manufacturing's back office is getting swarmed. The densest sub-pattern: AI agents for quoting, order entry, and procurement in US manufacturing. The interesting split is strategic, not technical. Arzana and Smartbase are building AI-native ERPs to *replace* the incumbent; Korso and Walter explicitly run *inside* customers' existing systems with human oversight. My read: sit-on-top wins the next two years — manufacturers don't rip out ERPs for startups — but the replacers are making the bigger bet, and Smartbase's narrowness (metal finishing shops, specifically) is the most honest go-to-market in the group.
ArzanaAI automation for the manufacturing office
KorsoThe Intelligence Layer for Manufacturing.
WalterAI Employee for Manufacturing Operations
SmartbaseAI-native ERP for manufacturers
2. The firm, not the software. A chunk of this cohort skipped selling tools to professionals and *became the professional*. Winter 2026 alone produced three AI-native law firms — Moritz, General Legal, and Vector Legal — all running the same play: AI does the drafting, credentialed lawyers sign, and the firm keeps the full fee instead of a license fee. Alchemize applies identical logic to customs brokerage, and Saudara AI to overseas sourcing. The margin math is seductive: you're not capped at the software line item, you take the entire services budget. The risk is equally obvious — you inherit a services business's operations, liability, and unit economics, with AI as your leverage rather than your product.
MoritzGlobal AI-native law firm handling legal work with same-day turnaround
General LegalThe AI native law firm for growth stage companies
Vector LegalThe premier AI-native law firm & legal operating system for Startups.
AlchemizeAI-Native Customs Brokerage
3. The adversarial-paperwork wedge. The sharpest wedge in the whole theme: workflows where a counterparty *owes your customer money* and the paperwork is the fight. Rote and Parrot both attack collision shops' insurance battles — denials, supplements, adjuster follow-up. Andco runs end-to-end case workups for personal-injury firms, and Wayco does the same for medlegal cases. ROI here isn't "save admin time," it's "recover revenue you were already losing" — the easiest pitch in vertical software, and the one place where citation-grounded LLM output is genuinely better than the outsourced back office it replaces.
RoteAI-native insurance department for auto body shops
ParrotAI-native operating system for auto shops
AndcoEnd-to-end case workups for personal injury firms.
WaycoAI operator for medlegal cases
4. Ops agents for the physical world, with zero new hardware. The last group runs operations for businesses whose work happens off-screen, with deployment friction engineered to near zero. transload turns the security cameras a warehouse already owns into a 3D freight-dimensioning system; Sidekick reaches deskless workers over plain SMS, no app install. Dayjob schedules skip-hire trucks through the booking and ERP systems waste firms already run, and Elyra answers the restaurant's existing phone line. The shared insight: in these industries the integration *is* the product, and "nothing new to install" is the moat.
transloadMeasure freight items with security cameras
SidekickAI-powered SMS assistant for deskless workers.
DayjobAI Scheduling for Short Haul Trucks
ElyraAI Reservation System for Restaurants
The cohort signal
This isn't a trickle — it's a program bet, and it's accelerating. Spring 2026 alone contains the entire manufacturing cluster plus the collision-shop, restaurant, trucking, and customs plays; Winter 2026 minted three AI-native law firms side by side; and Rote arriving in Winter 2027 shows the thesis still compounding a year later. Two consecutive batches stuffed with the same shape of company means YC has decided agentic vertical software is a repeatable template, not a one-off. The flip side: within-batch collision is now the norm. If your idea is "AI agents for [industry] back office," assume two batchmates are already building it.
Lessons from the last cycle
The prior cycle's verticals tell you which versions of this survive. Gusto and Zenefits chased the same SMB buyer; Gusto compounded quietly on payroll depth while Zenefits' growth-hack distribution ended in an acquisition — workflow depth beat clever channels. Flexport and ShipBob proved "be the service, not the software" can build giants — and also that it's capital-hungry and cyclical, while Frubana Inc going inactive shows what thin-margin full-stack looks like when it fails. Presto did voice AI for restaurants more than a decade early — reaching the public markets didn't save it from being early to a capability without owning the workflow around it. And Weave and Benchling carry the optimistic lesson: own one daily workflow (the front-desk phone, the lab notebook) deeply enough and you become the system of record by default.
If you're building here
Three openings I'd actually pursue this quarter:
- Run the Rote/Parrot playbook in an untaken trade. Adversarial insurance paperwork gates revenue in restoration, roofing, towing, and durable medical equipment exactly as it does in collision repair. You'd have to believe citation-grounded appeals measurably beat outsourced billing shops on win rate — and that you can prove it inside a 30-day pilot.
- Become the firm in a licensed profession that isn't law. Law has three batchmates; customs has Alchemize. Accounting, permitting and expediting, and insurance adjusting are still open. You'd have to believe you can carry the liability and hold services margins above software-investor tolerance while AI does the volume work.
- Niche down to one process industry, Smartbase-style. Pick a sub-vertical with one dominant legacy system and become its agent layer before going broad. You'd have to believe the TAM objection is wrong because you'll own 100% of a narrow workflow instead of 2% of a generic one.
The tarpits, by name: generic "AI ERP / AI agents for manufacturers" — Arzana, Korso, Walter, and qomplement shipped from a single batch, so entering now means out-executing four funded teams at undifferentiated positioning. AI restaurant phone and reservations, where Elyra is already live and the last cycle's drive-thru voice experiment spent a decade proving how brutal that buyer is. And AI-native startup law, three-deep in one batch. The pattern here is clear enough to copy — which is exactly why the only thing worth copying is the wedge, not the category.
Key companies in this memo
The headline bets — outcomes and all. (+14 more linked throughout the piece.)
RoteAI-native insurance department for auto body shops
ParrotAI-native operating system for auto shops
ArzanaAI automation for the manufacturing office
SmartbaseAI-native ERP for manufacturers
WalterAI Employee for Manufacturing Operations
KorsoThe Intelligence Layer for Manufacturing.
MoritzGlobal AI-native law firm handling legal work with same-day turnaround
AlchemizeAI-Native Customs Brokerage
ElyraAI Reservation System for Restaurants
DayjobAI Scheduling for Short Haul Trucks
AndcoEnd-to-end case workups for personal injury firms.
transloadMeasure freight items with security cameras
Build on this thesis
Generate grounded startup ideas steered by this memo — anchored to the real companies above.