Haptly

In late 2015, Nelson Shaw was freelancing and searching for a startup problem to solve. Spurred by the hype around drone applications, he and his co-founder began cold calling people about their drone usage. They discovered that dairy farmers in New Zealand faced a critical pain point: accurately measuring dry matter (grass levels) across their paddocks to budget feed for livestock. While farmers were manually "eyeballing" grass levels, no one had created an accurate solution using drone or satellite imagery. With Nelson's background in software and mathematics, and encouragement from local machine learning experts, they decided this was a solvable problem.

The team conducted extensive farm visits to understand the current dry matter estimation process and design a prototype. Nelson, the only developer, built a demo product using React and Material UI in about a month—learning React as he went. The demo showcased paddocks with dry matter totals and feed wedge charts. They presented it at a local farming trade show and received enthusiastic feedback from farmers, all eager to know when the product would be ready. This validation gave them confidence to move forward.

After joining the Vodafone Xone accelerator in early 2016, which provided $20,000 in funding, Nelson began collecting data from farmers and exploring satellite imagery as an alternative to drone camera data. Satellite imagery promised significant advantages: farmers wouldn't need to configure and fly drones themselves, and it would be easier to expand into countries with stricter drone regulations.

The team received strong validation through multiple channels. Accelerator acceptance brought press coverage that generated a constant stream of inquiries through their landing page. They also posted in farming Facebook groups, which attracted interest. Early test cases emerged—farmers willing to send data to help train their machine learning model. However, this validation proved to be a double-edged sword; looking back, Nelson felt they spent too much time on early validation when the market need was already clear.

What worked: market validation, accelerator support, and clear customer demand. What didn't work: the technical foundation. After 6 months of intensive development, Nelson hit a wall. Classical grass growth prediction models required more than just satellite imagery data (NDVI alone). They needed soil moisture, localized weather conditions, soil type, and grass variety information. At the time, only a handful of farms had all the necessary local sensors installed. The high cost of satellite data ($50k per year USD for New Zealand) compounded the problem.

By October 2016, after 10 months of work, Nelson couldn't see a technical path forward to build something meaningfully better than farmers' current eyeballing method. The biggest mistake, he later reflected, was spending too much time on early validation and not enough on determining technical feasibility upfront.

Haptly shut down in October 2016 without generating revenue. The $20,000 accelerator funding was spent on the business, and Nelson used savings and side income to live while working full-time. Importantly, neither Nelson nor his co-founder were deeply passionate about farming—they were "business/online software types" without personal attachment to the problem. Toward the end of Haptly, they began working on a different idea they felt more passionate about (later Contento, a guest posting platform). This shift, combined with the technical obstacles and lack of personal drive, led to the decision to shut down. Nelson has since shared his learnings: the biggest risk was determining whether the product could technically be achieved, and he would have spent more time identifying and addressing core business risks earlier in the process.