By FY26/4 Q1, Timee had become Japan's leading on-demand job platform, with 13.4 million registered workers, 440,000 registered client accounts, and a category it had helped define. After its 2024 listing on the Tokyo Stock Exchange, the growth question changed. Transaction volume alone was no longer the whole story. Timee needed to expand revenue per user and deepen the relationship between workers and businesses over time.

That opportunity sat in a specific gap. Timee Core was built for instant spot-work matching; Timee Career Plus, for full-time placement. Between them, a worker who had built a verified, trusted work history on Timee had no structured path to more stable, recurring employment. A business that had found workers it wanted back had no trusted channel to formalize the relationship. This was the gap Timee Crew was created to fill.

As the project's Design Director and strategy lead at Monstarlab, I was responsible for both the product and business strategy and the experience direction for Timee Crew, working directly with Timee's product leadership from concept through MVP.

*Timee Crew is the project name for what Timee publicly reports as its Long-Term Part-Time Hiring Support Service. As of the FY26/4 Q1 report and April 2026 survey disclosure, the service was being refined toward official release, with public validation figures drawn from Timee's published results and survey disclosures.

Timee Core's strength is its simplicity. A worker can accept a shift in minutes, with no resume or interview, and get paid the same day. That worked because the commitment was short and the decision was low-stakes. But as the worker base matured, a different kind of worker emerged: someone who had completed dozens of shifts, built a record of reliability, and was ready for something steadier. Timee had no product for them. Businesses faced the mirror image. Many had begun quietly identifying workers they wanted back, but the platform offered no way to formalize it.

The result was a structural gap. Timee captured transaction revenue at the spot-work end and high-value revenue at the full-time-placement end, but left the middle layer under-monetized and unserved: the repeat and stable hiring where earned trust turns into longer commitment. For Timee's most experienced, highest-trust workers, there was no reason to stay in the ecosystem as their ambitions grew. That made the gap a retention problem before it was a revenue one.

The first direction proposed by stakeholders was the efficient one: add longer-term placement features directly into Timee Core. It would reuse the existing user base, preserve familiar patterns, and minimize build cost. I argued against it.

Core's entire value rests on speed and minimal decision cost. Longer-term placement breaks that model. Workers have to weigh stability, workplace culture, schedule fit, and growth; businesses need stronger proof of reliability before committing. Folding that weight into Core would have made it harder to use for spot work and still inadequate for placement, weakening the product at both ends.

So I proposed the harder path: a standalone product with its own experience logic, but wired into Timee's shared infrastructure: One Timee ID, verified work history, payment records, and a path into Career Plus. The tradeoff was explicit and real. Extending Core was faster and cheaper to build; standalone was slower and more expensive. But standalone won on every dimension that mattered for the portfolio: protecting Core's simplicity, supporting high-consideration decisions, serving distinct worker intent, and scaling over the long term. Getting there required a direct conversation with Timee's product leadership about a risk that was easy to underweight: diluting the simplicity that made Core work in the first place.

Positioned between Core and Career Plus, Crew completes a single arc across the portfolio. Core generates trust through real, completed work. Crew converts that trust into recurring placement, grounded in verified history rather than a submitted application. Career Plus extends it into full-time career progression. For workers, that is a visible path from flexible earning to stable employment; for businesses, a trusted channel for repeat hiring; for Timee, an ARPU-expansion layer and a retention mechanism for its most valuable users.

To test whether that middle layer was commercially meaningful enough to justify a standalone build, I modeled the opportunity against Timee's installed base: adopting clients, recurring placements per client, monetization per placement, plus downstream value from Career Plus conversion. I deliberately kept the scope conservative: existing clients only, starting in food, restaurant, and retail; counting only placements converted from existing spot-work relationships; and excluding new-logo acquisition, new verticals, and any additional platform fees. Even on those assumptions, the base case pointed to meaningful incremental revenue before any of the upside levers, which is what justified treating Crew as a portfolio-level growth layer rather than a Core feature extension.

With the strategy agreed, I structured the Crew experience around four stages (Discover, Evaluate, Commit, and Manage), each built for the heavier decisions that longer-term placement requires. The stages are not independent steps but a cycle: each one feeds the next. Three principles held them together: continuity (Crew should feel familiar to anyone who has used Core), integration (it should connect seamlessly across the ecosystem), and progression (every action a worker takes in Crew should build toward their future, not just record their past).

Because Crew had to serve both existing Timee users and new ones, the experience had to be immediately recognizable while carrying weight Core never had to: stability, growth, schedule predictability, benefits, and workplace culture. With Timee's team, I led an expansion of the existing design system, adding components, a more flexible typography system spanning Japanese and Western needs, and a new illustration system to give the ecosystem more personality. To keep the cost of future expansion low, the team also built a proprietary tool that could generate new illustrations in the established style. The goal was not just the foundation for Crew but a reusable language across the ecosystem.

The first friction point in discovery was capturing what a worker actually wants from a longer-term role. A conventional onboarding survey asks people to articulate preferences they often have not formed yet. Instead, I proposed an adaptive system that starts from behavior. For workers with verified Timee history, the system infers an initial preference profile from past shift patterns, industry experience, and reliability signals; a short survey fills only the remaining gaps. New users move through an adaptive flow that adjusts to their answers. Either way, the worker starts matching from a position of relevance rather than a blank form, which shortens the time to a first meaningful application and helps both sides match faster.

There was pressure from senior stakeholders to put AI into the product, and the obvious instantiation was a Q&A chatbot. I pushed back: a bot answering fixed questions from a fixed playbook would add cost and confusion while doing little for either workers or businesses. Introducing AI without a clear role is how products get more complicated and less useful at the same time. The right question was where AI does something a rule cannot.

Evaluation was that place. Whether a job is a good long-term fit depends on more than matching stated preferences. It depends on a worker's trajectory, and on what a posting actually offers beyond its title. A fixed weighting algorithm struggles here, because the things that matter most (what a worker wants, and what a job really involves) are expressed in unstructured language, and no two job descriptions carry the same meaning even under a shared template. That gap between structure and meaning is exactly where AI earns its place. I defined the matching model as a contextual analyst: it combines the preferences gathered in Discover, the worker's past behavior, and Timee's historical data from Core and its partners to assess how a given posting maps to a worker's desired growth path, and to surface that reasoning rather than just a score.

Applying can be intimidating when every employer's process is different. I proposed a standardized application built from two flows. The fixed flow (basic information, resume, commuting details) draws on One Timee ID, surfaced inside Crew as Timee Resume, so workers edit and confirm rather than re-enter. The variable flow (employer-specific questions, the worker's motivation for applying) is where real effort belongs, and where AI assists so the whole application can be completed in well under the time a from-scratch process would demand. The principle is simple: spend the worker's attention on what is genuinely theirs to decide, and automate the rest.

Workers in longer-term roles frequently hold more than one placement at once, which makes conflicts and paperwork the real day-to-day burden. Using Timee's API to synchronize schedules across partners, I designed Crew to detect potential schedule conflicts early and prompt quick remedies before they become problems. The Manage surface also gives each job a set of fast, common actions and generates the documents these arrangements require (tax certifications and other legal paperwork), so the administrative weight of stability does not fall back onto the worker.

The same discipline shaped the second AI role. To meet Timee's full range of users, including workers in their late 50s and 60s who make up a meaningful share of the base and struggled with the standard flows, a fixed-script assistant that answered set questions and escalated to support would have fallen short. Building on the clear structure of the four stages, I defined Ask Timee as a semi-agentic experience: a companion that can guide and actually complete steps alongside the user (finding roles when someone does not yet have a clear picture, or moving an application forward) rather than answering and handing off. The result is an assistant that lowers the floor for less confident users without narrowing what it can do for everyone else.

In 2025 we released the first Crew MVP to selected clients in food, restaurant, and retail, verticals where Core already had deep penetration and where the move from spot work to recurring roles was most natural. I structured validation around three specific assumptions rather than general feedback: that workers would see Crew as meaningfully distinct from Core; that businesses would treat verified work history as a more reliable hiring signal than their informal workarounds; and that a standalone-but-connected architecture would let people move across the ecosystem without confusion.

The first two held, and Timee's published results bear them out. Of the workers whose preferences Timee captured in-app, roughly 700,000 signaled interest in moving toward long-term roles, confirming the demand the gap implied. More than 60% of businesses that approached workers through the service successfully converted them into long-term hires, and 56.5% reported that long-term hiring through Timee lowered turnover, direct evidence for the verified-history thesis. Just as important for the original strategy, the pilot showed no sign that converting spot workers into longer-term roles cannibalized spot-work revenue; on the company's own account, early data pointed to increased Timee usage frequency. The architecture bet of expanding the portfolio without weakening Core was holding.

The third assumption was where the honest finding lived. Ecosystem fluidity showed early promise but exposed friction in the handoff between Core and Crew. That result did not undercut the thesis so much as locate the next problem to solve, and it directly shaped the following development phase.

Crew added a structural layer Timee's ecosystem had been missing, a dedicated path from verified work experience into longer-term placement, without compromising the simplicity that made Core work. By keeping the products architecturally distinct but experientially connected, the portfolio can now serve three different intentions (immediate work, recurring placement, and career development) inside one shared ecosystem.

The harder lesson was about influence at the platform level. Building separately instead of extending was not the obvious call, and making it required turning a design argument into a business case and earning alignment across two organizations. An MVP with externally validated assumptions, a conservative commercial model, and early signs of broader client interest was the evidence that the right architecture, argued clearly, can take a platform further than the path of least resistance would have.

Lastly, I would like to offer my sincerest gratitude to the teams at Monstarlab and Timee. Timee Crew would not have been a possible achievement without the dedication of each and every members participated in the project.