In the demo, the co‑founder of Manus AI shows three examples of Manus’ use cases, one of which I will deep‑dive into one of them, property research in New York State. In this example, they show Manus conducting research to filter properties in New York based on criteria such as low crime rates and high‑quality schools for the user’s kindergarten and high‑school children, for a family with a combined monthly income of $50K. Manus AI first breaks the task down into three steps: the research phase, where it identifies neighbourhoods based on safety, school quality, and budget; the search phase, where it uses Redfin (not sure why it specifically chose this website) to create a comparative analysis; and the reporting phase, where it prepares a property‑comparison report for the end user. It begins by opening a browser and searching for news articles reporting crime rates in New York neighbourhoods. Then Manus carefully reads multiple articles to identify middle and high schools and writes a Python program to calculate the budget the user could spend on purchasing their new apartment. Next, Manus fires up the browser agent again to look for listings on Redfin in the recommended neighbourhoods, such as the Upper West Side, within a $1.8M to $2.6M budget ( calculated from the monthly income but with certain assumptions in mind ). Finally, Manus gathers all insights and writes a detailed report on its recommendations, including the key benefits for the user’s family for each property. All of this in minutes!

Now, for simplicity assume you are the product manager of this tool to help with property research, and are responsible for the success of this tool in the market. We assume the success criteria to be the number of final property purchases made by the user that your tool had suggested. This would include secondary metrics such as how many sessions did it take for the user to come up with the right suggestion, and user satisfaction scores. A few questions that come to my mind immediately are —

1. How do we know if a user went through a happy flow? or an unhappy flow? Yes - thumbs up, thumbs down and rating are brute force but how many of us actually rate it? If no, what implicit signals (click‑through rate, rapid refinements) do we use to judge feedback?

2. At what point do we start the task execution vs ask the user for the assumptions they may have implicitly made? For example in this use case we do not know the other financial obligations or current lifestyle of the user. Their willingness to spend on their new home and location could change accordingly. More importantly criteria such as pet ownership may seem trivial for the user but is actually very important during the search process. One approach could be to come up with the assumptions and ask upfront, however, that could lead to a delay in the “aha moment” for the user, so what assumptions are optimal to ask beforehand vs letting the user iterate later?

3. Manus writes a python program to come up with a potential budget. At this point would it make sense to pause ask the user for their financial obligations and then move to the next step? Or would it make sense to just move ahead with the assumptions assuming the user has shared the important information in the initial prompt. What would be the right financial model to actually come up with these estimates? Should we give this lever to the user?

4. Next you need to work with the GTM team to roll this product out to the right users, how would you test the edge cases, regional questions, etc of this tool to make sure it actually provides value to my ICP? How would the overall user conversion funnel look like?

Well the answer to most of these questions lie in the iterative cycle of product development, by A/B testing each of these core assumptions against our KPI. While tools like Mixpanel, Amplitude, exist to answer such questions for web applications, the “paths” in AI Agents are not clickstreams, but conversation trees which are complex to track. The construction of a conversation tree starts with the first prompt given by the user to the AI Agent, followed by the visual cues generated as a part of the “deep thinking” shown to the user by the Agent. Followed by the final output given by the agent to the user. The conversation tree additionally includes the response from the agent to each subsequent query of the user. This conversational tree, once discretized for each user interaction via a combination of computer vision and NLP algorithms, would need to be analysed for implicit signals of the user impacting the primary business KPIs.

Once you come up with the possible issues the user is facing, another mammoth lies in updating the AI agent to handle the unhappy flows. Every time you tweak a model or a prompt you need confidence that your KPI will improve! In the SaaS era, changes in one part of the flow would most likely not affect the user experience in a completely different flow of the user, so testing was relatively faster, however, now any update to the model or prompt could change the user flow ( aka user experience ) in a completely differently functionality. We would need a set of evals to test all the flows after any changes in the model or prompt, major or minor. Think about it like notion-type templates of evals that we run every time to ensure the KPI looks better at-least during offline evaluation. But are eval framework provided by the current tools such as Arize AI, ML Flow, etc good enough to test an ensemble of models? How do we actually know multiple prompts are not interacting to create an unhappy flow? Do we actually need a simulation bot to interact with our product wearing different hats of the user? A simulation bot that would try to intentionally break one of the user flows pushing the product to the edge?

If you're building AI agents and facing these challenges, let's design the future of AI product analytics together. Feel free to reach out here.

The shift with Agentic AI systems is that they now become the user interface, rather than just a backend enabler like traditional machine learning systems. Earlier, fixing a bug or usability issue was a simple code change—typically resolved in an hour, or at most a day, and you were good to go. But with AI agents acting as the interface, incorrectly classified intents or poorly handled flows require either retraining (or fine-tuning) the entire AI model, or modifying the prompt. Either way, you must re-test all user flows, since they're all driven by the same underlying model. This can take days or even months, depending on your use case and by then, user trust is already lost.

Real-world Example

While building our symptom checker chatbot at Zealth, we developed an entire knowledge graph of symptoms and their corresponding home management protocols. We also built an LLM that could understand the user’s intent, extract symptoms, ask follow-up questions, and suggest the most relevant management protocol, such as diet or exercise recommendations, for home care. We had accounted for symptom related queries like “I am feeling sick“ or “What should I eat because I’m feeling nausea.” For non symptom related queries, such as “Zyada protein ki cheeze batao,” we provided a standard out-of-context response in the first version. However, as soon as we went live, the very first user query was “Sir k pichle wale hisse me dard hai” ( “Sar” also sometimes spelled as “Sir” in hindi means “head” and the sentence actually means “Pain in back side of the head”). But our chatbot confused “Sir” in english with “Sar” ( or head ) in hindi from this sentence & the conversation went in a totally different direction—leading to a poor user experience. The issue stemmed from how we had structured our prompts. In hindsight, we should have handled such inputs by asking follow-up questions. But in our rush to release the first version and learn from it, we hadn’t accounted for this edge case during testing.

Unlike traditional software testing which has definite number of pathways in our product / feature, AI agents have infinite number of pathways making it impossible to test them all!

High Stakes & Real Consequences

A user hitting an unhappy flow can be extremely costly, especially in high-stakes applications. For example, in our case, the first patient of a doctor we had been pursuing for almost a year, who also happened to be a close friend of the doctor, encountered such a scenario. They discussed it with the doctor, and as a result, we lost the doctor’s trust and the pilot opportunity. This one bug taught us more about real-world agent UX than 200 happy-path evals ever could. Similar situations in other healthcare or finance-related bots highlight the need for exhaustive testing of boundary cases. It's critical to ensure that the bot does not make diagnoses or provide recommendations that are non-compliant with industry standards—right from day one.

Why does this problem exist?

Current evaluations rely on manually reviewing user queries and selecting test cases, a process that’s time-consuming and limited by team bandwidth. At Zealth, we started by manually chatting with patients on WhatsApp to gather real conversations. From 200 such chats, we created our training and test datasets. But with only 30 conversations in the final test set, we barely scratched the surface of real-world scenarios.

And it’s not just us. This challenge resonates across the industry. In general, the most common way to assess the usability of AI agents is by setting up evaluations (evals). The most popular LLM eval tools, such as MLflow, DeepEval, and Arize AI, focus only on evaluating individual models or prompts within an agentic system. As one Lead ML Engineer at a major transport company in SEA put it: “Testing individual ML models is easy. Testing how they interact in a real product is the hard part.”

The biggest risk isn’t broken flows. It’s misunderstood users. Current eval tools miss emergent UX failures that happen when multiple agents, models, or prompts interact in unpredictable ways.

How to solve?

This problem can be reframed from a data science perspective—what if you could simulate all possible interactions with your chatbot, based on user personas learned from their digital behaviour, in just a matter of minutes? You could then evaluate these simulated flows against your key performance indicators (KPIs), automatically flagging unhappy paths or risky interactions, allowing you to fix them before real users ever encounter them. Such an approach could even enable proactive testing for bad actors and edge cases, including scenarios where non-compliance or unsafe recommendations might occur, critical for high-stakes domains like healthcare or finance.

If you're tackling similar challenges or exploring solutions in this space, I'd be excited to exchange ideas. Let’s connect.

As a starting point, we began tracking our sleep every night using our Amazfit watch, which was usually lying around discharged before this. While it was fun to see our sleep time, wake-up time, deep sleep duration, and sleep stages, we weren’t sure what to do with the data. Aside from calling each other out for not going to bed by 11 PM or sympathising when one of us didn’t get enough deep or REM sleep, we didn’t know how to actually improve the quality of our sleep further.

In our pursuit of finding a reasonable answer, we tried various solutions and products we found on the internet. We realized that we were only tracking our sleep on nights when we felt good before going to bed. However, on nights when we struggled to sleep due to illness or overthinking, the first thing we did was remove our watch and place it on the side table, as we weren’t used to sleeping with a watch on our wrists. It felt more like a hindrance than a help on the days we needed it most! We also tried an under-the-mattress sleep tracker similar to Withings, only to notice a very similar pattern. Maybe it was the sleep performance anxiety of knowing we were going to get bad data on nights we already felt bad? We’re not sure. 🤔

Objective vs. Subjective Sleep Tracking

While the smartwatch provided us with some objective data, we also wanted to track subjective factors from the evening before sleep — such as caffeine intake, chamomile tea consumption, any medications taken, and, most importantly, whether we had traveled during the day. To do this, we bought a sleep journal, which we diligently filled in on good days but found ourselves avoiding on bad days. Maybe sleep tracking could only be maintained consistently if we saw some improvements on a regular basis. With this in mind, we decided to explore other solutions that claimed to enhance sleep.

Trying Out White Noise and Micro-Current Stimulation

• White Noise — We realized that white and pink noise have helped many people fall asleep faster and experience fewer awakenings due to fewer disturbances during the night. Excited by this, we got a device specifically for white noise, only to find it irritating — probably because we weren’t used to it. Later, I learned from my friends that it works well for some people, especially babies, as they are familiar with similar sounds from the womb.

• Micro-current Simulation — Some devices claim to use microcurrent stimulation to influence the central nervous system, potentially promoting the production of melatonin and endorphins, which are involved in sleep regulation and mood enhancement. These devices may aim to modulate brain activity, including strengthening delta waves, which are associated with deep sleep. While some studies suggest that electrical stimulation can impact brainwave activity and sleep quality, the exact mechanisms and effectiveness of microcurrent devices for improving sleep are still being researched and are not fully established. Personally, when we used this device, the microcurrent pulses had the opposite effect, awakening us when we were dozing off to sleep! 🤨

Learning from Others’ Sleep Journeys

We also decided to explore the experiences of our friends, as well as others on LinkedIn and Reddit, who may have tried devices or solutions to optimize sleep. We spoke to around 50 people about their sleep journeys, trying to understand what had helped them — or not helped them — improve their sleep. We found a few common patterns!

1. 60% of the people who recovered from a phase of insomnia did so primarily by making changes and strictly adhering to sleep hygiene practices. Specifically, they went to sleep regularly at a fixed time, avoided meals at least 3 hours before bedtime, avoided screen time at least 1–2 hours before going to bed, and maintained a cool bedroom environment. That’s it! Each of them had read the book Why We Sleep by Matthew Walker.

2. 25% of the people who were experiencing sleep issues mentioned that meditation, particularly breathing exercises just before bedtime, helped them sleep better when overthinking was the cause of insomnia. This was mainly during times of stress, including experiences like becoming new parents, job loss, or illness in the family. Personally, I also found that meditation helped me calm my mind, making it easier to fall asleep.

3. While supplements such as melatonin, ashwagandha, and magnesium helped some of them, only 5% had tried taking supplements for sleep. Additionally, visiting a general practitioner for sleep issues did not lead to any significant outcomes.

4. None of them knew exactly what to do with the sleep stage data from their Apple Watch, Whoop, or Ultrahuman Ring. 🤨

Final Thoughts

It’s interesting that sleep hygiene and breathing exercises proved to be the most effective ways to enhance sleep quality. Currently, most sleep technologies are focused solely on capturing sleep data. I’m eager to see how technology can make it easier to incorporate these habits into everyday life. If you are working in this space, I would love to brainstorm.

Before the sleep workshop, my understanding of sleep was mostly limited to getting more deep sleep, and dreams occur during REM sleep! I only realized during the workshop that I was completely wrong.

Polysomnography is the gold standard for sleep stage classification, whereby light sleep (Stages N1 & N2), deep sleep (Stage N3), and REM are supposed to be approximately 55%, 20%, and 25%. More deep sleep will lower the amount of REM sleep; hence, having more of any one of these stages might not necessarily mean better sleep. Moreover, these smartwatches and smart rings “estimate” the sleep stages based on pulse and heart rate variability. As per the sleep specialists (pulmonologists, neurologists, etc.), sleep quality cannot be objectively measured like glucose levels or BP. Instead, the best sleep is when you are “fully alert” in the morning, have few to no awakenings during the night, and experience no daytime sleepiness, primarily measured through widely accepted questionnaires such as the PSQI sleep scale.

As a technology person, I went through Amazon to order a ton of sleep products, including sleep (melatonin) gummies, bed cooling systems (Eight Sleep or other alternatives available on Alibaba.com) and a wide range of trackers such as watches, rings, and mattress trackers. I only realized during the workshop that the most effective approach to improving sleep quality is primarily through better sleep hygiene — consistency in lights-off time and waking time, ensuring screen time is minimal after 9 PM so that blue light does not disrupt your circadian rhythm, and avoiding late-night munching or having an early dinner to put your intestines to rest by bedtime. Additionally, I had the myth that snoring is a sign of “sound sleep,” only to realize it is one of the symptoms of a larger underlying condition — obstructive sleep apnea — which requires clinical intervention.

While we all know that sleep improves memory and cognitive functions — spindles was an important concept that amazed me. Sleep spindles are brief, high-frequency brain wave patterns that occur during deep sleep, as shown in the image below.

Sleep spindles occur as part of the brain’s process of consolidating memories from the day, primarily moving information from the hippocampus to the neocortex (parts of our brain) for long-term storage. Additionally, they help with processing emotionally salient memories. There are a lot of interesting studies on spindles leading to interesting conclusions, such as their positive correlation with IQ. In children, high spindle density during the N2 sleep stage supports rapid learning and neurodevelopment. Spindle activity is also used as a biomarker for aging and other conditions such as schizophrenia, MCI, Alzheimer’s, depression, and epilepsy. Surprisingly enough, the characteristics of sleep spindles are so unique to an individual that they can be used as an electrophysiological fingerprint and is genetically inheritable.

While learning about sleep was one aspect of the workshop, talking to neuroscientists working on dream cognition at NIMHANS was eye-opening — specifically inducing lucid dreaming (dreams where you are self-aware that you are in a dream). During childhood, I used to induce lucid dreams unknowingly, and I enjoyed it a lot since I got complete control of my dreams, particularly jumping around the school like Superman! However, these dreams did not last long enough as I would end up waking up. What is interesting is that lucid dreaming is used as a possible therapy to treat PTSD by recreating the scene and completing the unfinished deed. This phenomenon dates back to the Senoi tribe, who had a culture of discussing and sharing dreams to understand an individual’s thought process, which is believed to be the reason they had some of the best mental health.

A few studies also showed that voice or smell-based stimuli during sleep (particularly during the deep sleep phase) could help with treating addictions as well as improving memory consolidation. In one example, a study showed that smokers who received a putrid smell treatment during their deep sleep phase had a 30% chance of reducing smoking in the following week. During the deep sleep phase, these participants were given a putrid smell along with the smell of a cigarette to ensure their brain associated the smell of a cigarette with the putrid smell. In the following week, 30% of the people who received this treatment reduced smoking. In another example, a study showed that auditory stimuli during the deeper phase of sleep improved the ability to solve cognitive tasks amongst youth. During the day, these subjects were given a light background sound while solving cognitive tasks, and the same light sound was played during the deeper phase of sleep when memory consolidation happens. It turns out these people had an improved ability to solve these tasks in comparison to the control group.

While devices like Eight Sleep and Oura Ring offer valuable insights into sleep tracking, there remains an untapped potential to focus on enhancing sleep quality and its impact on cognitive functions. This gap presents an exciting opportunity to create solutions that truly improve quality of life through sleep. If you share this interest or have ideas to contribute, I’d love to connect and discuss potential solutions. Please reach out through this link to schedule a time to talk!

The first day at Skydive Spain (SDS) was the ground school where I was taught a lot about the aircraft, equipments, about the different parts of skydiving — freefall, canopy control, landing procedures and most importantly safety procedures. The instructors at SDS put a lot of focus on repeating the procedures via simulations again and again on ground. After a certain point my “over-confidence” kicked in and it seemed silly to repeat the procedures so many times on the ground. Little did I realise how important it is, till after my first dive. During freefall, most of the things you do is instinctive, much like driving a car. Most of the actions are performed by muscle memory, while the conscious brain is busy processing a new environment. It’s only then I realised the power of repetition and the impact our sub-conscious mind can have on our actions.

The following day, I woke up early and looked at the skies. It was blue with almost no clouds, making me both excited and nervous in anticipation! My level of nervousness constantly kept increasing as I wore my rig (container with parachutes), moved to the boarding area, and climbed up into the airplane. Going up to 15000ft takes almost 15–20 mins in the airplane. On the ride up, naturally my mind tried to get more worked up, but I could see a visible difference on the face of AFF students as compared to experienced skydivers in the plane. The experienced skydivers were super excited, happy and energetic all the way up to 15000 ft — talks a lot about how the experience is going to be!

At approximately 12000 ft, we did a final gear check (It’s called 3x3 check). At 15000 ft, the speed of the plane is lowered, the pilot gives a green signal and the door opens. Cold air rushes in and I realised this is the moment! The experienced skydivers (who plan to pull their parachute at a lower altitude) jumped out first, followed by which a couple of AFF students jumped and I knew it is my turn! Naturally my heart-beat started rising. My instructor, Davide, showed me the hand signal to breath and relax and I moved to the edge of the plane. I stood on the door looking at the clouds below and our planet Earth :P. I could feel the wind gushing past my face and the entire body! I looked at the wing of the airplane, started exit procedures of checking in for readiness with both my instructors …… and took the leap of faith …..

For the first 2–3 seconds, my mind was processing what just happened, followed by which I got (somewhat ?!) stable flying on my belly. All the feelings of fear and anticipation evaporated instantly and I realised this is the most beautiful feeling I have ever experienced! I was fully aware of my surroundings and flying in the air! As per the plan, I had completed three practice pulls and checked my altitude. I was at ~9000ft knowing that I still had some time before I would pull my parachute (planned at 6000 ft). I started looking at the horizon enjoying the breathtaking views of the puffy clouds. But guess what?! within 5 seconds my instructor shows me the hand signal of “watch your altimeter” and I have already dropped to 6500ft! Wow! That was when I realised how fast we lost altitude during a freefall.

At 6000 ft, I opened my parachute and both my instructors tracked away to keep a safe distance. That was the first time I realised I am solely responsible for myself 5000 ft above the ground. The words of my instructor from ground school rang in my ears “It’s you under your canopy! 🤷🏻‍♂️” As per the training, I quickly took control of the toggles, observed my surroundings, moved left & right to check the movement of parachute. I oriented myself relative to the ground and moved towards the drop-zone. I had another 5 minutes before starting the landing pattern at 1500 ft. Under the canopy I took some time to soak up the scenery — the beautiful houses, fancy cars moving on the tar and muddy roads and the airplane (I had jumped from) landing on the runway below (! not directly below me). I followed the landing pattern (which is like a highway of parachuters) and safely landed on the ground with some help from my instructors over the radio. The busy mixture of emotions in the last 15 minutes made me feel supremely lifted and uber relaxed on the ground.

The following levels of the AFF course increased my hold on skydiving on a stepwise basis.

• In level 2 — I was required to do a left and a right turn during free-fall.

• Level 3 was the most interesting — as this was the first time both my instructors left me and I was supposed to free-fall on your my own maintaining stability.

• Level 4 onwards I had only one instructor, Saray, with me during my free-fall. In level 4 — I was supposed to turn 90 degrees left and right on my own.

• Level 5 is similar to level 4 where I had to turn 360 degrees both sides this time.

• By the time I reached level 6 — I already had a hold of aircraft exit procedures, my mind was much more relaxed while going up in the airplane, my free-fall stability was much better, mind was trained to look at the altimeter very frequently, my canopy control and landing as well was much smoother.

• The last level (level 7) is the most interesting, as this time I was on my own from exit of airplane to landing, my instructor, Saray, was with me during the jump but she only observed from a distance. In this jump they certify for solo jump as a skydiver!

After completing all the 7 levels of AFF, and paying my beer fines to everyone in the dropzone, I was ready for solo jumps. I had to complete at-least 10 consolidation jumps ( solo jumps ) and AFF level 8 (hop and pop) before getting my BPA A-licence. The first solo jump was a bit scary I must say! Going up in the airplane I realised this is the first time I am jumping without any instructor! I convinced myself that I had been trained well for this and I quickly revised everything for my jump, from the exit, free-fall position, what to do if I get unstable during free-fall, pulling the parachute at the right altitude to emergency procedures. As soon as I left the airplane — it felt complete freedom! I just kept observing the ground, the horizon and kept changing body positions during the freefall!

By the time I was doing my consolidation jump 2 & 3 — the process of wearing my rig, doing the 3x3 check and going up in the airplane to 15000 ft was a very fulfilling and happy experience. I realised I grew exponentially personally and as a skydiver, from where I was 2 weeks back.

Just when I got used to jumping from 15000ft & started missing the adrenaline rush, there is a new exciting challenge to complete — the AFF level 8. This level is designed to give a feel of exiting the airplane at a low altitude as a skydiver. You remember, I used to pull our parachute at 6000 ft during the AFF course? This time I had to jump out of the airplane at 5500ft, wait for 3 seconds and pull my parachute not more than 10 seconds after jumping. It gave a whole new experience to skydiving, where I could visibly see the ground rushing towards me for those few seconds, and after I opened the parachute I could see the airplane disappearing into the clouds at almost my eye-level. Skydiving gave me views from a totally different angle — ones that were beyond my imagination.

A total of 21 jumps down the line, I got my BPA A license! Yaay! 😅 I can’t be more than excited to go to the next step and learn formation skydiving! This entire experience was much more than skydiving! The process of going from 0 dives to 21 dives made me grow as a person!

1. I realised, with some perseverance and grit, it is possible to conquer any fear and the rewards are enormous with a sense of accomplishment! More specifically, I realised I could be aware and follow instructions under extreme circumstances!

2. The view from above the clouds falling towards the earth gave me a perspective of how small each one of us is in this enormous planet! It literally and figuratively brings you down to earth.

3. Skydivers come from all walks of life! Different countries, beliefs, age, etc all with one goal of enjoying the jumps in a safe and responsible manner. Becoming friends with such a varied set of people, with the same passion, gives you the ability to embrace different perspectives.

4. It is therapeutic — the positive afterglow from skydive has a lasting ripple effect on other aspects of life. In a parallel universe, if Zealth’s office was next to a dropzone, I would go for a skydive every morning before starting my day!

Statutory Warning : Skydiving could be addictive and therefore injurious to your financial health.

This irrational decision coupled with my curiosity to observe and experience life under the waters, motivated to book my spot for training at Temple Adventures in Pondicherry!

The course starts with a theoretical lesson — explaining the basics of underwater pressure and buoyancy. My first day of training was in a swimming pool. The first breath under-water through your regulator (mouth-piece through which you breath) is sensational! All my life, I went into panic mode as soon as I was submerged underwater in a swimming pool or a bath tub, but here, after the initial reaction — being comfortable while completely under the water was a new feeling altogether.

As a part of the training in pool & lake, we were taught some key skills that had to be demonstrated to our trainer. The ones I really liked were —

1. Mask removal 🥽 — While you get comfort in the fact that you will be wearing a mask underwater to prevent water touching your eyes or nose. This skill requires you to remove your mask, ensure you don’t panic before you can put it back and blow air to clear the water from the mask. The key here was to make sure you consciously breath through your mouth and not involuntarily from your nose.

2. Hovering🧘🏻‍♂️🧘🏼‍♀️ — I personally liked this skill since it advocated mindfulness! When we breath underwater — air fills in the lungs and increases our buoyancy, therefore moving us upwards towards the surface, similarly we move downwards towards the bottom when breathing out. In this skill, we had to control our breathing to stay consistent at a depth.

Each of these skills were simulations of what to do — and be comfortable in the ocean!

The first day in the ocean was nothing like I have experienced before — It was a bright sunny day with very good visibility underwater. Observing the aquatic life underwater while floating was much more calming that I could imagine. 😇 The first school of fishes was batfish🐠 — followed by gun fish🐟 & we were lucky to spot some sea horse near the reef as well! It was so calming that I could just stay there and keep watching for the entire dive, but Rajat (my dive instructor) & I had to goto the bottom and perform some skills on navigation. What amazed me underwater was the fact that complex sentences such as “Move 10 flaps (1 flap ~ 1 meter) towards the north and come back to the current point following a square pattern”, was seamlessly communicated to a newbie like me using only hand signals . This was the first time I appreciated the language of hand-signals!👌🏻☝🏻🤙🏻

While the experience was enjoyable — I admit, I did have my “O Shit” moments during the dives — nothing scary but it was just a part of me still getting comfortable underwater

1. Buoyancy control in the sea — During my first dive in the ocean, buoyancy control was all that I was trying to achieve during the dive — now looking back it was funny enough since my instructor, Rajat, was stable near the sea-bed while I was going up almost 2 meters while inhaling and rapidly dropping on the sea-bed during exhaling — this continued in cycles. Thanks to his guidance underwater and during the debrief post-dive, I was much better in the second dive.

2. Yeah, I was guilty of disturbing aquatic life underwater during my first dive. In the mission to control by movement underwater, I temporarily lost control and swam right into a school of fish which dispersed immediately probably thinking I was threat to them. It was fun to watch, but is dangerous and frowned upon in the dive community.

3. Swim test — Okay this was scary even before I started, standing on the edge of the lake (yeah not just the deep swimming pool I am scared of), I was preparing myself to go 100 meters towards the center and come back. Ignoring all thoughts of sinking to the bottom at the center of the lake I started swimming towards the center, slowly observing the bottom disappear while trying hard to stay positive. Well — while it sounds all good, I admit I had panicked when I had to turn around in the center but I knew I had no option left other than swimming back to the boundary of the lake! Thank god Rajat didn’t make me do it again! 🙈

Apart from the fun of observing life underwater — the open water course taught me some key life-skills of staying calm in an alien environment even when unexpected things happen. All my instructors were super-cool which made the experience more fun & am super excited to complete the advanced open water course as well.

“A story has no beginning or end: arbitrarily one chooses that moment of experience from which to look back or from which to look ahead” — Graham Greene

In July 2019, I took the hard decision of leaving the comfort zone of a job with a fixed monthly salary to move to a new country and start my entrepreneurial journey. Now that I look back, it felt similar to the initial hesitation of discomfort when entering a cold shower, or walking over the ledge just before my bungee jump. Once I went past it, I enjoyed the ride so much that I never regret taking the plunge. I was lucky to have cushion while managing my financial and family dependencies to join, Entrepreneur First (EF), in Singapore. EF is a talent investor which brings in 100s of solo founders every batch — all with one objective of finding their co-founder, pitching their team and idea to an investment committee, getting a pre-seed fund, and starting a company 3 months after joining. While leaving my last job, in my head, I assumed the ideal scenario of getting funded and starting a company by November 2019, but life had other plans!

The kick-off weekend at EF gave an adequate flavor of what was coming next. Having been a data scientist for the most parts of my career — putting on my headphones and coding or brainstorming with a close-knit team was a typical productive day. In contrast — the objective of the kick-off weekend was to identify my potential co-founder, after getting acquainted with more than 80 new people within two days, that came from 23 countries. Discussing with a diverse set of people ranging from pilots to PhDs, to discussing how AI could help discover the next best perfume was exhausting but exhilarating, to say the least.

Coming from an archetypal Indian family & being trained to be a good guy since childhood — it was difficult to accept the necessity to be borderline unethical during customer validation. In order to validate an idea in the healthcare space, we had to barge into chambers of doctors without an appointment, or use the infamous technique of sending personalized mass emails to doctors showing interest to collaborate with them, or surprise the doctors with our interview questions instead of reporting symptoms of illness after booking a consultation with them. Yes, most of the time we were thrown-out politely but the few insightful discussions were worth it!

I teamed up with Monika, a Ph.D. in Immunology, complementing each other while working on an idea in cancer care. We had good validation of our problem and a letter of intent from multiple doctors across 5 countries to use our solution once it was built. Our mentors at EF were super-confident of us & rated us one of the best teams in terms of progress & validation. We pitched our idea to the global panel of EF — only to get a shocking phone call two days later from our mentor breaking the news that we had not been funded by EF because it was a conflict of interest with one of their parallelly funded companies!

Before the pitch — we were naively over-confident of getting funded after our venture partners appreciated us for our seemingly flawless idea and pitch. From our presumptions funding & starting a company— to not being funded, customers being skeptical for not having the support of any organization & not having an office space was a huge blow to our conviction & confidence. We had reached out to many investors outside EF only to be turned down since we were too early in the journey for angels.

By this point, uncertainty was slaughtering us — but, we learned the most important lessons of our ride during that time

1. We learned the importance of criticism more than ever — the more we got criticism from a wide range of investors, customers, and peers, the more we learned and became stronger. It has reached a point where we ignore positive comments and reach out to people who are most likely to criticize us.

2. We learned to proactively keep our emotions on a check — being too happy, or too sad at any moment clouded our judgments, and you read an example of that in the above paragraph. It has reached a point where we proactively take time out to come out of any of the extremes of emotions before taking up the next task.

By December 2019 — we took a small break to visit home. Once we were back, we acknowledged our shortcomings and decided to bootstrap for 3 months based on our individual finances. We used this time to give our best on building the MVP and getting feedback on our thought process from as many experts as possible. By January 2020, we had pivoted based on the discussions, had a decent MVP, and got a positive response from multiple doctors we had reached out to in various geographies. This gave us confidence that we were on the right track. We were lucky to have Antler, the largest tech startup generator, like us as a team and accept us into the deep tech executive program starting in February 2020.

None of us know how the end is going to look like — but the journey to the end will never change and we learned the importance of enjoying every bit of it. In January 2020 while building the MVP, I took the liberty to enjoy a workacation in Chiang Mai, one of my best solo trips in recent times. Additionally, we visited a lot of doctors in February 2020 but made sure we treated ourselves with pani-puri and dabeli in Chowpatty even after being literally trashed by the doctors, or had a sumptuous dinner at Mannat, a punjabi dhaba, after having a positive conversation.

By the end of March 2020, we had pitched to the investment committee at Antler, a tech-startup generator, and were fortunate to have a pre-seed from them. We are excited to have Antler’s support us in our journey ahead to build Zealth-AI. There are lots of challenges coming ahead, with the most recent one being the black swan of 2020, COVID-19! We aim to come out of this tsunami stronger while building Zealth-AI. Whatever it is, will be a story for another time.

Huge thanks to all our peers, friends, and family to help us go through the ups and downs. Special thanks to our mentors at EF and Antler for believing in us.

Please feel free to reach out to Monika or me to help us move a step ahead in achieving the vision of building solutions for personalized cancer care by connecting us to any doctor/hospital in the field.

I was both excited and nervous while landing at Chiang Mai, as I didn’t know what to expect from this place. I was gladly received by Ladislav and Urana, the owners of Amaka (the hostel I had booked). Well, I had actually asked for an airport pickup service while booking my hostel, but the fact that they came to pick me up themselves while talking about Chiang Mai on the way back, made me feel like a home away from home.

The hostel was a pleasant surprise! We had dorm-mates from multiple nationalities, everyone was in Chiang Mai for a unique purpose. They ranged from a 25-year-old Taiwanese girl taking a break from work to a 70-year-old Japanese guy traveling across the globe as a part of his retirement. Each one of their stories was stimulating.

Each city has its special story and Chiang Mai has its own! The old city is guarded by walls that were built to prevent invasion. It had 8 gates, each having its own astrological significance and unique purpose. Eg. The Phae Gate faces eastwards, signifies prosperity and is used by the monks and traders to enter Chiang Mai. I enjoyed listening to more such stories while talking to the local people — be it while hiking to the chilly Doi Inthanon, or visiting Doi Suthep to view the sunset.

Contemporary Chiang Mai has its own set of activities to offer — be it relaxing your mind through Tibetian bowl meditation at Chiang Mai Holistic, or joining random groups for trivia night at cafe bistro, or relaxing your muscles with a Thai massage, or getting your spontaneous creativity to action at Improv, or partying hard with people from across the world at Zoe in yellow! The list goes on and on — it’s time that restricted me to try numerous other activities!

Solo trips are addictive — mainly because you get the opportunity to explore yourself while being the director of your plan! There are plenty of things I had a chance of enjoying since I was traveling solo — be it learning about cannabis cultivation from a Californian at trivia night, going on a dinner with a group of French people I had randomly met during the hike, trying out the variety of local cuisines northern Thailand had to offer or discussing for 30 mins with a barber on the different styles of beard I could experiment with! Each of these was a peculiar learning experience for me.

Well it’s no fun if you don’t work hard while partying hard! Chiang Mai has its unique attributes to aid you to work hard. There are plenty of co-working spaces that offer a daily pass. I had tried multiple co-working spaces in Chiang Mai. The best ones were -

1. CAMP at Maya mall — it’s a funky co-working space open 24 hours. It has a balcony facing Doi Suthep mountains. The view of sunrise from the mountains at 6:30 in the morning is good enough to pull an all-nighter while working at CAMP! 🌄

2. MANA — Very quiet place with comfortable chairs. They offer chilled water and tea throughout the day to keep you hydrated (actually over-hydrated 🤔) while working.

Would I go to Chiang Mai again if I need a break while working? Yes!

Would I recommend visiting Chiang Mai? Of course yes!