Most founders entering the healthcare space chase clinical breakthroughs like new diagnostics, novel treatments, better patient-facing apps. Far fewer turn their attention to the back office, where massive operations teams spend their days shuffling schedules, verifying licences, and managing shift changes so that doctors can actually see patients.
Joe Shearman and Jack Light went after the paperwork. After working across engineering, forecasting, and workforce software, including an earlier scheduling startup that scaled before hitting a ceiling, they zeroed in on the operational burden that keeps clinics running but patients waiting.
Now, as co-founders of Planbase, a Y Combinator–backed AI platform managing scheduling for healthcare providers, Shearman and Light are building the infrastructure to take that weight off.
Two Paths to the Same Problem
Shearman and Light came at the same question from opposite directions: how much of a company's day-to-day operations can be handled by systems rather than people.
For Shearman, that started at McLaren Automotive, where he worked as a launch engineer responsible for getting cars into production. In practice, much of the job involved processing change requests and working through enterprise software like SAP to get updates approved and pushed through.
"You'd think you spend most of your time actually getting things out the door, but honestly, most of the time you spend is just clicking buttons," Shearman says.
The work exposed how much of execution depends on navigating internal systems and processes, rather than making decisions. Getting anything done required moving through layers of approvals, tooling, and coordination.
Light was looking at the same type of work from the opposite angle. During his research in econometrics at the University of Chicago, he focused on demand forecasting and optimal worker allocation—how to determine how many people a business needs, where, and when. The goal was to take operational data and turn it into clear staffing decisions, replacing manual planning with something more systematic.
Together, the two perspectives pointed at the same gap: even if you can model the "right" decision, organisations still struggle to execute it through fragmented systems and manual workflows.
Shearman later moved into workforce software at Workforce.com, working across product, solutions engineering, and implementation. That gave him a closer view of how scheduling and staffing decisions are actually made and carried out in practice.
Between them, they had both sides of the problem: how to decide what should happen, and how to make it happen reliably.
Entering the Startup World
Shearman and Light started looking for an opportunity to commercialise Light's doctoral research on demand forecasting and optimal worker allocation. They first founded Elando AI, a startup that predicted staffing needs for fast-food chains, including Domino's and Popeyes. The product would ingest sales data (like how many pizzas or garlic breads a location sold last week) to forecast demand and recommend how many delivery drivers to schedule and where to route them.
By most startup metrics, Elando AI was working, landing multiple large customers. But the co-founders hit a ceiling they couldn't engineer their way past. The business model required selling through third-party platforms, meaning customer acquisition, the product roadmap, and even the underlying data architecture were all controlled by partners.
"Customer acquisition is controlled by the platforms you plug into, which, as a founder, isn't particularly fun; it's like selling through Walmart or CVS," Shearman explains.
The product, they realised, was functionally a feature of someone else's software instead of a standalone platform. And the technical limitations were just as constraining: to build a scheduling algorithm that actually worked well, the team needed to own the entire software stack, from the data model to the user interface. Outsourcing any of those layers to a partner degraded the quality of the output.
Founding Planbase to Solve Healthcare's Operational Bottlenecks
After stepping away from Elando AI, Shearman and Light started looking more closely at where operational coordination breaks down in high-stakes environments.
In conversations with healthcare companies, a consistent pattern emerged. Much of the work behind care delivery is still handled manually, spread across fragmented systems and internal processes. Tasks like coordinating shifts, managing availability, verifying licences, and handling last-minute changes often sit across multiple tools, with no single system responsible for keeping everything aligned.
Healthcare is costly, and clinicians command high salaries, so every misallocation of labour carries steep costs in both directions. Overstaffing increases payroll, while understaffing limits patient access to care and creates gaps in coverage that are difficult to recover from in real time.
The administrative burden behind these workflows is significant. Shearman estimates that for every doctor, there are multiple administrators handling tasks like moving shifts, verifying compliance, and tracking licences. Much of that work exists to bridge gaps between systems rather than to make new decisions.
Existing workforce tools, they found, were largely general-purpose. They were designed to serve restaurants, retail, and hospitals with the same underlying systems. Healthcare's licensing requirements, regulatory constraints, and uneven demand patterns make those tools difficult to apply directly without significant manual coordination.
In late 2023, Shearman and Light started building Planbase, an AI workforce platform designed specifically for healthcare providers. Instead of producing a single static schedule, the system uses AI agents to continuously handle the operational work in between—forecasting demand, coordinating clinician availability, filling gaps, and monitoring compliance across state licensing requirements.
As the product took shape, the founders entered Y Combinator's Summer 2024 batch and secured their first enterprise customer before the programme began. That deployment expanded from an initial group of clinicians to a broader rollout over time.
Building by Listening
In working with healthcare teams, Shearman and Light found that scheduling rarely follows a single plan. The process involves too many variables and too much situational context for a single automated output to be reliable. In practice, decisions often rely on operational knowledge held by roles like practice managers or chief nursing officers.
Planbase reflects that by treating scheduling as an ongoing process rather than a one-time decision. Its AI agents handle tasks like contacting clinicians about availability, coordinating replacements when someone calls out, and monitoring compliance across state licensing requirements.
Shearman attributes the company's traction to a simple discipline: letting customers dictate what gets built next rather than chasing trends or investor narratives.
"Forget VC hype or what's trending. Just see what customers want and build that," he says.
After a decade in workforce management spanning a traditional scheduling platform, a one-shot optimisation engine, and now a conversational AI approach, Joe Shearman and Jack Light's goal is straightforward. They want to strip out the work that sits between patients and the doctors they need to see. "Our goal is to remove the administrative overhead around care delivery and let clinicians focus on patients, rather than navigating systems," Shearman explains.
ⓒ 2026 TECHTIMES.com All rights reserved. Do not reproduce without permission.
