18 months of testing Posha - Our Journey towards Product Market Fit

When we first set out to test Posha, it wasn’t just any kitchen gadget.

It was something that had to cook, think and adapt in real time. Unlike traditional appliances, Posha had to make real cooking decisions, adjusting for heat, texture, and timing just like a home chef would.

In short - we weren’t building a cooking robot, we were teaching a robot to cook. And the best way to test that? Inside real kitchens, with real people, making real meals.

For 18+ months, Posha lived in homes across different households - tackling different recipes and feeding different families.

Our main goal with this was to figure out: how many times a week were people using Posha for their meals? Were people willing to alter their cooking behaviour?

This is the story of how we approached our testing phase and how we went from Posha — an exciting concept to Posha — an everyday consumer hardware product that fed people food.

Cooking Up a Storm in the Bay Area

As a Bay Area startup, we had to keep our initial testing close to home — ensuring that our early users were within reach.

But finding the right users wasn’t as simple as putting a call out for beta testers. It was a completely new concept for someone who had never seen or heard about Posha. And it was hard to visualize or imagine. They had to watch Posha in action and try the food to become a believer!

We hosted “On The Road” events, rented a space, set up live demonstrations, and ran ads to invite people to taste robot-cooked meals.

And they showed up. We had some great conversations and live demonstrations and got some good feedback. Some of those we met at these events continue to be customers to date!

We set up booths at various events, using Posha’s live cooking as a curiosity magnet to draw people in. Watching a robot in action was enough to spark interest, giving us the perfect chance to strike up conversations, learn about their cooking habits, and invite them to join our beta program.

Laying the Table

Early on, we made a critical decision—people had to pay for these trials. Why?

Because true, unfiltered feedback only comes when users have skin in the game. When something is free, people tend to be polite or indifferent. But when they pay, they care. They engage. They tell you exactly what they love, what frustrates them, and what needs to change.

So, we set the price at $50/month for Posha, allowing users to pause and resume their subscription as needed.

For a product that had been in R&D for what felt like forever, this was our first real revenue—and hearing that Stripe notification go kaching was a feeling like no other.

The First Sizzle

We started with just five devices in the field and gradually scaled up to fifty.

From day one, we were deeply hands-on with the entire process.

Each user began their journey with a 45-minute onboarding call, during which we learned about their eating habits—what they cooked, how often, and their preferences.

Instead of relying on shipping partners, we personally delivered every single device. No unboxing surprises, no guesswork. We spent an hour—sometimes more—walking each household through the setup, showing them how to use Posha, setting expectations, and even sharing a few cooking tips to get them started.

We tried to make their experience as frictionless as possible. We supplied pre-measured hand-spices for free. We kept extra devices and spare parts on deck. We also built a 24/7 support team that was available via phone and chat, ensuring help was always handy.

Trials by Fire

Despite all this, as people started using Posha, the journey wasn’t exactly smooth sailing.

New challenges kept cropping up.

We had people issues: Some users wanted to cook larger portions, especially when guests were over. Others experimented with ingredients they weren’t supposed to use. Many preferred cooking with what they already had instead of buying specific ingredients for a recipe.

We had hardware issues: Some functions were not working smoothly at times- oil wasn’t dispensing properly, stirrer was getting stuck. The camera would get fogged up, thus blocking Posha from cooking the recipe successfully.

Challenges = Playbook

Every issue was a chance to learn.

We could’ve just shipped a spare part if a user needed it. Instead, we showed up in person, turning every visit into a deeper conversation about their experience.

Our manufacturing and technical teams were based in Bangalore, while our business team operated out of the U.S. This meant that when a technical issue arose, our business team needed to handle it on-site.

To prepare them, we ran a one-week crash course on Posha’s inner workings. They learned everything—from troubleshooting the chip and replacing the motor to, if needed, even soldering parts themselves.

Food is deeply personal. How people cook, what they eat, and when they eat is shaped by their family structure, home environment, work schedules, and more.

By being there—inside their kitchens, sharing conversations —we gained invaluable context. We were not building a cooking robot; we had to build a product that truly fit into people’s lives.

Stirring Up Insights

Hardware

• Cooking extra portions
  • Our robots could cook 1-4 servings per session, but we noticed most users— even in two-person households—consistently cooked four. They weren’t just making dinner, they were planning for leftovers. Soon, users, especially four-person households, complained that our pan wasn’t big enough for both dinner and extra portions. Our Beta prototype had fixed capacity limits, but in our next hardware iteration, we significantly increased the pan and ingredient container sizes—while maintaining the same product footprint.
• Cleaning > Cooking
  • In hindsight, this seems obvious, but it wasn’t to us then. One of the biggest factors in kitchen product adoption is ease of cleaning—sometimes even more important than its primary function. Realizing this, we redesigned our next hardware iteration with completely flat surfaces, minimal crevices, and dishwasher-friendly parts.
• Had to fit below the countertop
  • The kitchen countertop space is sacred. We learned our product had to be low enough to fit under any U.S. kitchen cabinet and compact enough for users to keep it on their countertops daily.
    

Food

• Home > Restaurant
  • We had wrongly assumed that people would be excited to use a cooking robot to recreate restaurant-quality meals at home. So, when we first designed our menu, we packed it with dishes that felt more like something you’d order at a nice restaurant rather than cook yourself. But as users started cooking with the robot, their feedback told a different story. They weren’t looking for fancy, restaurant-style meals—they wanted simple, home-cooked dishes, the kind they ate every day. And more importantly, they wanted those dishes made exactly the way they were used to—just like they would cook them, with all the little personal touches that made them feel familiar and comforting.
• Not just recipes - Sandwich fillings, sides
  • We initially assumed people would use the robot to cook full, end-to-end meals. But our users had other ideas—they were way more creative than we expected. Soon, we started getting requests for all sorts of things beyond just full meals. People wanted fillings for sandwiches, burritos, tacos, and quesadillas. Others asked for sides—like sauces and roasts—to complement their main dishes. It became clear that our users weren’t just looking for complete meals; they wanted the flexibility to mix, match, and integrate the robot into their own cooking routines.
• All processing stages
  • It wasn’t just about choosing different recipes—we realized that even within a single recipe, users wanted flexibility in how they cooked. They didn’t always want to start from scratch; sometimes, they preferred to use previously semi-cooked or processed ingredients. Take Penne Arrabbiata, for example. Some users wanted fresh tomatoes, while others preferred canned crushed tomatoes or bottled pasta sauce. Everyone had their own way of making the dish, and they wanted the robot to adapt to their preferences, not the other way around.

Software

Autonomy software often stumbles when it first meets real users—ours was no exception.

• Users rarely did what we expected.
• Ingredients meant to be prepped a certain way weren’t.
• Quantities were off.
• Sometimes, ingredients were skipped entirely—or swapped for something completely different.
• Users improvised, and our AI struggled to keep up.

We used it as a starting point.

Every unexpected action became a data point. We continuously expanded our training dataset, capturing the full spectrum of real-world cooking behavior.

Our North-Star

Our main north star metric throughout was: Are people coming back and using the product? Because at the end of the day that's what really matters.

The software, the hardware, the food and everything else had to come together for that single pursuit.

As we made progress month-over-month - three things made us realize that we were close:

• The hardware was reliable and met the standard we had set for ourselves.
• Our recipes were consistently rated good and people were enjoying them.
• People were continuing to use the product while paying for it.

People are creatures of habit. We’ve cooked a certain way our entire lives. We’ve built systems and routines around it.

If a product like Posha has led to behaviour change in the kitchen, has changed the way families fix their everyday meals and transformed their mealtime rituals, it's not a small feat.

It's something that's very, very special to us and nothing short of a revolution.