Pivoting from Agtech to Data Automation: Challenges and Learnings

I originally joined YC's Startup School in 2019 with a project in Agtech, named Kulti.

The original verbatim goal was "Affordable automation Agtech for sm-mid farmers in the developing world".

In practice, the idea was to produce a Philips Hue-like ecosystem of low-cost sensors to drive an automation-based revolution, eventually powered by cognitive services, and concretely, the first product was a water sensor connected to a cloud service with a custom alert system.

In the beginning, one of my largest motivators for the project was a strong belief that I had the following unique advantages:

  • The ability to work on the entire stack from the electronics and the microcontroller of the hardware, all the way to the cloud infra and the frontend and the services in between.
  • Privileged access to a large and connected user base of farmers, who did emit a strong interest signal in such a solution
  • A pragmatic target market, who believes in investment and ROI, and used to middle to long term dynamics
  • Experience in Product Design, and specifically in Human / Automation interactions

And beyond all of this, I was extremely motivated by the small but eventually dramatic impact of better water management for the climate, and for the economics of developing world.

I got a hardware + software prototype working, and iterated on it twice after a few pilots, with relative success, but ultimately I had to put the project on a hiatus.

Why is that? Some of the challenges were exciting and can be overcome, but others are just a complete hindrance.

  • The supply chain for hardware components is a nightmare of complexity and slowness. Once you stray away from your off-the-shelf ESP8266, you have to rely on suppliers from China and the delivery times (1-2 months) mean your iteration cycle is measured in quarters instead of weeks (when you don't receive a faulty component lot and have to order it again).
  • The logistics to ship prototypes, go on site to evaluate the conditions and hangout with the users too complex for an urbanite, especially when my ideal users live in other countries.
  • I discovered many subtleties to the problem, and the specialist knowledge required (in soil engineering and botanics specifically) slowed down my progress significantly (but it was amazing to learn how plants actually work).
  • The amount of moving parts and technologies needed for a solution covering custom-designed hardware and a cloud solution with a mobile clients is just mind blowing, and very complex operationally

I parked this project for the time being and moved on, but the learnings were immensely valuable. Many of these learnings might be duh moments, or already known in the industry, but sometimes you have to fail to learn first-hand :-)

If I could summarize for the benefit of others, basically:

  • For hardware projects, live near your suppliers or in an industrial hub (I considered briefly relocating to Shenzhen)
  • In the same vein, make sure your customers are within a couple of hours at worst. This is not a problem if you don't need to iterate on the hardware solution (but who doesn't ?)
  • Software engineering and design skills are necessary but very far from sufficient, and while you might not get away without it, nowadays it's not a decisive success factor
  • There is only so much you can do on your own, and if the initial problem is not "easy" enough, your throughput is an incredible limiting factor that impacts your motivation first before everything else
  • If you must ship a complex solution, try to isolate a slice where you can shine and deliver maximal value, and find partners for everything else.

So what am I doing now?

I've launched https://monitoro.xyz since a few weeks, and it's growing steadily. Stay tuned for a follow up post to talk about the transition in more details.

  1. 2

    Thank you for your post! It's encouraging to see another hardware-oriented person on the forum. I am in the midst of building theautomatedfarmer.com on the side. I have all my prototyping materials so now I just need to get my word out!

    Best of luck!

    1. 2

      Thank you @jawmes.
      I just looked at your website, looks like a great initiative! Just signed up :-)

      1. 1

        Thanks man! I didn't see your subscription come through via MailChimp so it seems I may have to make it easier to sign up? I think there is a confirmation email that's detracting others too.

        Anyway, thank you for your encouragement! Can't wait to share what I've been building!

        1. 2

          Yes! I did receive a confirmation email that I missed :D

          Consider my email yours! haha

  2. 1

    Wow, what an amazing story. Im just curious:

    1. Why didnt you just continue developing the product with off the shelf hardware until you reached product market fit?
    2. How were you acquiring customers?
    3. What was your process for getting feedback from them given your customers?
    1. 1

      Hey Scott thank you for your great questions!

      1- This one might be hard to answer without diving into the technicalities, but trying to keep it high level:
      There's already a large offering of small garden water sensors you can buy anywhere half-decent, that my project was strictly inferior to in the first iteration (which did use in fact components I could get the same day on Amazon).

      After performing some research in botanics and soil engineering to understand where these products are lacking when it comes to agriculture in opposition to gardening, I realized I needed much more precise (and different) measurements than just the water level itself.

      It turns out my actual competition was farmers calling in an actual soil engineer on-site who would be performing some sampling of the soil and send it to a lab for further analysis (which could take days or weeks and costs a leg), as well as perform regular check-ups.

      While it looks like an exciting disruption opportunity (buy these little devices, no need to call in an engineer anymore) and it is, the technology to actual perform that kind of analysis on the spot is much more cutting edge and is not readily available in your friendly neighbourhood electronics shop. I had a few bad experiences trying to source it and after doing the math, it was clear this project's pace would be too slow for a bootstrapped one man start-up project (done on the side!)

      2- I actually reached out to some extended family members who are in the target demographic, and asked them to introduce the concept, with my prototype at hand.
      Since farmers are pretty well networked (and they almost all source typical farming supplies from limited locations), it was relatively easy to achieve a pretty large outreach

      3- Phone calls, a couple of (very sunny) on-sites. Very low-tech in this regard.

      1. 1

        Thats interesting, were your sensors designed so that you would not need a soil engineer to take samples? Is this process to complicated to automate (perhaps thats the product to develop)?

        1. 2

          I did come up with a design for the product that should replace a lot of what the engineer is doing based on my research and calculations, but I did not get to complete it at the time I put the project on hold.

          So it was not tested, but yes the goal was definitely to replace this slow sampling+analysis process.

          It's not overly complicated per se, it just needs calibration and a lot of resulting trial and error.

          1. 1

            Gotchya, I think Teralytic is tackling this problem as well, although I can't say for sure if they're trying to make the process plug-n-play. I have had a lot of interest in AgTech for years, and always have an interest in hearing about startups in this space. Thanks for sharing!

            1. 1

              you’re very much on point !

              Teralytic targets the exact same technical problem, but:
              1- Their pricing is much much above what the developing world could afford (sm mid farmers not big ag)
              2- The engagement model with farmers is via a leasing model (which might not be a bad thing)... BUT
              3- They’re mostly available in North America

              So it looked to me that there’s room for a new tiny player to grow :D

Trending on Indie Hackers
Hit 🎉 40K users 🎉 ahead of the launch 15 comments The one where writing books is not really a good idea 13 comments I build an awesome white-label product but I don't know who could need it. 9 comments 99 Designs and 2 Dribbble Designers Later I Have a Logo! 7 comments I got 3 offers to buy my product for $100k, $125k and $150k, rejected them and here's why it was the correct call 6 comments Straight No to Bootstrap after Tailwind! Is it only me? 5 comments