Mike Rondinaro
Co-founder
Tech that scales: why Parkday has to feel fast a week before lunch
There’s a cartoon of a Thanksgiving table where every guest has a different label:
Vegan
Vegetarian
Pescatarian
Low carb
Dairy free
Gluten free
Severe allergies
It used to be a joke. Now it looks like a normal office headcount list.
Now stretch that table to 150 or 1,000 people, over five workdays, across multiple floors. Everyone has different health needs, preferences, and cultural expectations. The restaurants all have different menus, capacities, and prep constraints.
That’s the problem Parkday’s tech is built to solve—not just at 11:59am, but a week in advance.
We’re not optimizing for a single frantic decision window. We’re optimizing for a planning window: menus drop, people plan their week, restaurants get real signal early, and the actual lunch rush becomes execution rather than panic.
From one-off private chefs to planning lunch for the whole office
Before Parkday, we started with a very different product.
One of our first companies, Umama, matched people with private chefs who would cook in your home. It was magical when it worked—truly personalized, restaurant-grade meals built around one household.
But as a business, it had a problem: every experience was bespoke, 1:1, and hard to scale.
When Parkday was founded, we realized the same personalization instincts would be far more powerful in a different setting:
What’s a better problem for software to solve?
– Designing the perfect menu for one person every once in a while…
– Or designing the right mix of meals for hundreds of people, five days a week, from all the local restaurants available to them?
The second problem has more constraints, more data, more impact—and a much better fit for software.
The twist: to do it well, you can’t just think about “what’s for lunch today?” You have to think in weeks.
When do people see options?
When do they decide?
When does the restaurant need to know?
How do you avoid the “Doordash at my weakest moment” spiral our Evolution of food access piece talks about?
High-frequency and high-foresight
Most food apps are optimized for in-the-moment ordering:
You’re hungry.
You open the app.
You scroll a giant marketplace.
You pick something that looks good right now.
That design has consequences:
It hits you when you’re tired, stressed, or rushed.
It makes it harder to stick to longer-term health goals.
It gives restaurants almost no runway to plan, staffing-wise or inventory-wise.
Parkday flips that—but not by swinging to the opposite extreme.
We still live in the real world of meetings, cravings, calendar changes, and surprise celebrations. Our goal isn’t to shame impulse or force asceticism. Instead, we design a curated system that leans toward:
health, based on your own personal definition
independent restaurants and local supply chains
variety across cuisines and formats over the course of a week
Most alternatives fall into one of two buckets:
“Slop” style programs that optimize for ultra-cheap, lowest-common-denominator food with little personalization.
Programs with impractically early cutoffs that technically reduce uncertainty but don’t match how modern teams actually live and work.
We’re very intentionally in neither camp.
Menus drop weekly, not hourly
For each office, menus go live about a week in advance:
You can see your choices for the whole week in one view.
You can plan lunches the same way you plan workouts or meetings.
You can “set it and forget it,” with room to adjust if your schedule changes.
We intentionally designed the app to nudge decisions earlier:
Weekly view that makes it natural to pick ahead.
Clear cutoffs (for example, 9am same-day) so expectations are obvious.
Gentle prompts that remind you, “Hey, your Thursday and Friday lunches aren’t set yet.”
The result is a pattern we want:
A big chunk of orders placed days in advance.
A meaningful wave the night before.
A natural bump up to the 9am cutoff for last-minute deciders and changes (with enough morning runway to still remove your order if your day shifts, which meaningfully reduces food waste through carefully designed timelines).
We absolutely still engineer for that morning bump—but our primary optimization target is the shape of the entire week, not a single spike.
Because most ordering follows a predictable weekly pattern, our systems make it immediately obvious when something is off for a specific office or service. Our ops team can jump in, understand the why, communicate with clients, and adjust menus, messaging, or timing. That level of visibility into a client’s food behavior simply doesn’t exist in open marketplaces or old-school catering, which mostly see disconnected, one-off events.
Every so often we see a piece of feedback that says “Sunday drops feel intense.” We actually designed them that way on purpose: as a weekly cultural moment that gets people geared up and excited for the week ahead. That said, timing and cadence are easy to tune, and we regularly tweak them to match each company’s culture and internal rhythms.
Why this is better for restaurants
For restaurants, earlier decisions are gold:
They can staff correctly and order the right ingredients.
They can prep intelligently instead of guessing.
They can say “yes” to bigger corporate volumes without chaos.
Because we know our users deeply and longitudinally, we can predict demand with meaningful accuracy at both the topline and meal level. That prediction gets better when people lock in earlier.
Instead of calling a restaurant at 10am and asking them to suddenly serve 200 lunches, we’re sending them a strong signal days ahead and a very accurate signal hours ahead.
Why this is better for people
For employees, weekly planning is quietly powerful:
When you think about your week of eating, you tend to set higher goals for yourself.
You’re more likely to say, “I’ll go big on Friday, but keep it lighter Monday–Thursday.”
You avoid the “scroll until I’m seduced by the worst possible choice” dynamic that comes with marketplaces.
Our anti-marketplace thesis is simple: marketplaces catch you in the moment at your weakest and don’t help you stick to what you wanted. Curated, planned menus help align actual behavior with intentions.
This was one of the original reasons behind our founding and vision: giving people private-chef-level curation without requiring a private chef. It’s the same logic that explains why far fewer people need a private car than need a ride from A to B via Uber, but thoughtful food guidance has long-term value far beyond a commoditized trip from A to B.
Performance where it matters: week view and workday
Weekly planning doesn’t remove the need for performance during the workday—it raises the bar.
There are really two critical performance surfaces:
The planning surface
Weekly menus with tons of nested data loading instantly
Personalization applied across days (not just “today”)
Easy bulk actions (for example, “lock in Mon–Thu, tweak Friday”)
The live-service surface
Real-time ETAs and pickup windows
Clear flags for which section to grab your meal from
Fast updates when there’s a change (delay, swap, etc.)
Browsing for future services while you’re waiting for today’s lunch
We treat both as first-class citizens. Under the hood:
We pre-compute user-level recommendations and aggregate them at the service and week level.
Our APIs are intentionally designed and tuned: some endpoints are big, data-dense workhorses because performance and prediction require it, but they’re highly optimized, cached where it makes sense, and continuously monitored so the app can still render “your whole week” quickly.
We monitor latency, memory usage, error rates, outlier queries, and more aggressively around both the menu-drop window and the pre-cutoff morning bump.
In other words: the system has to feel light when you’re calmly planning Sunday night and when you’re checking your pickup flag at 12:05pm.
Designing the UI to pull decisions upstream
We didn’t just hope people would plan ahead—we designed for it.
Some of the intentional choices:
Week-at-a-glance layout
You see all your upcoming services in one scroll, not buried behind a calendar picker. It feels natural to tap through and choose meals for the week in one session.Clear “done-ness”
Days you’ve already locked in feel visually “complete.” Empty days feel obviously incomplete—just enough friction to remind you.“Default to your best self” patterns
It’s easier to pick a whole week of meals aligned with your personal definition of healthy, then make one or two fun edits later, than to win the decision battle five separate days in a row.Graceful last-minute changes
Life happens. You can still make morning-of tweaks and cancellations up to the cutoff, and the system and restaurants are built to absorb that.
This is the opposite of the “infinite scroll of everything near you” pattern. It’s more like a great playlist you made on Sunday that carries you through the week—curated, intentional, and easy to live with.
A small team, a large surface area
Parkday’s tech surface area is bigger than it looks from the home screen:
Personalized recommendations at the ingredient and meal level
Weekly planning logic across offices, vendors, and services
Restaurant capacity planning and prep guidance
Delivery routing and timing
Nutrition data and integrations (like Apple Health macro sync)
We handle all of that with a relatively small engineering and product team. That’s on purpose:
The people who write the code are close to restaurant ops, to nutrition, and to customer success.
When someone reports, “The week view feels sluggish for our SF office,” it doesn’t vanish into a ticket graveyard.
We can ship improvements that respond to real-world food and logistics constraints, not just abstract metrics.
We built a lot of this before AI dev tools were everywhere, which forced us to get the fundamentals right: clean data models, clear service boundaries, and observability. New AI tooling is an accelerant on top of that, not a crutch.
Tying it back to our core thesis: from marketplaces to curation
In The evolution of food access, we argued that food has moved from scarcity to abundance—and that the next step has to be curation.
Marketplaces gave everyone a “button” to press for food at any time. That solved access, but it didn’t solve decision quality. In fact, it often made it worse.
Parkday’s weekly planning model is one of the concrete ways we operationalize that thesis:
We don’t show you everything; we show you a curated set of meals from restaurants we vet deeply for sourcing, nutrition, and quality (see also From the Farms and How Parkday finds restaurants).
We don’t demand five separate “what’s for lunch” heroics from you every week; we help you make one thoughtful plan and then glide on that.
We don’t ask restaurants to react to chaos; we give them signal and lead time so they can do their best work.
The tech is only “scalable” if it respects the very real constraints of kitchens, delivery routes, office life, and human willpower. Weekly planning is the bridge between all of those.
Why this matters for companies
For our clients, this isn’t just a UX flourish. It shows up in outcomes:
Better restaurant relationships. Predictable, well-signaled volume means better service, better pricing, and access to higher-end partners who wouldn’t touch chaotic corporate orders otherwise.
Better employee health and satisfaction. A week of intentional, aligned meals beats five days of “whatever’s at the top of the app.” Over time, that compounds.
Better operations. Office managers aren’t stuck running ad-hoc headcount every morning. They can glance at the week and feel confident.
Because the experience is genuinely pleasant—people tell us that ordering on Parkday is a bright spot in their day—it also builds a cultural habit:
“We eat well here.”
Quick FAQ
Do orders still spike before the daily cutoff?
Yes, we still see (and design for) a bump leading up to the 9am cutoff: last-minute deciders, schedule changes, cancellations. But it’s not the whole story—and it’s not the behavior we optimize the product to encourage.
Our UI, notifications, and weekly planning tools intentionally pull a large share of decisions earlier in the week, where they do the most good for both restaurants and people’s long-term goals.
If people plan a week ahead, can they still change their mind?
Absolutely. You can edit or cancel orders up to cutoff. Weekly planning is about defaulting to your best intentions, not locking you into a rigid schedule.
How does this all relate to “high-frequency ordering”?
We think of it as high-frequency, high-foresight ordering:
High-frequency: Most employees interact with Parkday multiple times a week.
High-foresight: Many of those interactions are planning or tweaking, not scrambling.
That combination is what drives better outcomes for restaurants, employees, and companies.
