Not screens: we build the record and the action

Plask, the company behind Laney, was founded in 2020 by three people who studied engineering at POSTECH. It began with NAVER D2SF's campus startup program in 2019; the AI we built after founding won a CES 2023 Innovation Award. We were named to Forbes Asia 100 to Watch, CEO Junho Lee and CPO Jaejun Yoo made Forbes 30 Under 30 Asia, and we spoke at NVIDIA GTC the same year.

Once the technology was recognized, we asked where it was needed most. The answer was the most complex operation in Korea: the clinic. To receive one patient, website, phone, messengers, booking books and ad tools run separately, and the front desk stitches them together by hand. Laney put all of it on one data model and passed 100 customers within 10 months of launch.

Laney is five people today. Five of us can serve 100 customers because we work with AI ourselves: the product, sales material, customer reports and this careers page are all built with AI. We use every day what we sell.

For twenty years B2B software sold screens. People bought a CRM for its dashboards and pipeline views, priced per seat. Those screens made staff enter data, and muscle memory made switching hard.

Now AI agents read and write data directly, without screens. a16z calls it software "losing its head"; Salesforce launching a headless, API-first product is part of the same shift. Meanwhile anyone can build the tools they need with AI, and more companies cancel subscriptions saying "we can build this ourselves." Software that only sells screens and features gets harder to keep.

Both pieces land on the same answer: the screen may go, but these matter more.

• The system of record: the one set of data a business trusts. The more agents there are, the more they need the same record.
• Rules and permissions: the unwritten way a business works, plus which agent may do what on whose behalf, with an audit trail.
• Owning the action: products that act, capture the outcome and use it to decide better next time.
• Data the product creates: not imported data, but data that exists only because the product is in the loop.
• The last mile in the real world: businesses tied to execution that won't be fully automated.
• Buyers who can't build it: operationally complex businesses without engineers won't build and maintain this themselves.

And the software that survives doesn't tell customers to adapt to it. It becomes the base on which what each customer needs can be built quickly.

Laney is the customer record for offline businesses and the AI that works on top of it. Against the criteria above:

Clinics are the hardest test among offline businesses: high-value customers, long consultations, strict rules and many unwritten ones. If the record-and-action loop runs here, the same frame runs elsewhere.

The sign on the door differs; the frame is the same. Customers ask, consult, book, visit, pay and come back. Only each industry's rules and tone change: a dermatology treatment becomes a pilates class, a vet's vaccination cycle, a law firm's consultation slot.

Beauty and aesthetics, fitness and pilates, vets, academies, law and tax offices, auto repair, home services, lodging: every business that meets customers in person and runs on bookings has the same problem, complex operations and no engineers.

1. Deeper in clinics: close the loop from first inquiry through visit and return.
2. More industries: keep the data frame, swap the rules, and move beyond clinics.
3. A door for agents: prepare for customers' own AI assistants to ask and book through Laney, with Laney as the trust layer that decides and records what each agent may do.
4. Network: benchmarks by industry and region and links between businesses, so each new business makes Laney more useful to the rest.
5. Execution and payments: deposits, payments and dispatch, closing the loop into the real world.

These are examples of directions we want to explore, not things we've built. In the FDPO assignment you can pick one or propose something better.

• a16z — Is software losing its head? ↗
• Namanyay Goel — AI is Killing B2B SaaS ↗