Tule
Summer 2014AcquiredHelp farmers make irrigation decisions with sensors / computer vision
Pressure-test this opportunity
Explore the risks and possibilities with a prompt for ChatGPT, Claude, or your agent.
Help farmers make irrigation decisions with sensors / computer vision
Explore the risks and possibilities with a prompt for ChatGPT, Claude, or your agent.
Tule Technologies was a precision irrigation startup founded in 2013 by Tom Shapland, a UC Davis viticulture and agronomy PhD, and later joined by co-founder Jeff LaBarge. The company built the only commercially available hardware sensor for measuring Actual Evapotranspiration (ETa) — the real water use of a farm field — and participated in Y Combinator's Summer 2014 batch. Based in Davis, California, Tule expanded from hardware into software (FieldStat) and eventually AI (Tule Vision, a computer vision app for measuring vine water stress), serving high-value specialty crop growers in California's vineyards and tree nut orchards.[1][2]
Tule's failure as a venture-backed company was not a product failure — it was a capital structure failure. The company built scientifically rigorous, hardware-dependent technology that was too capital-intensive to scale to venture returns in a niche market. When the next funding round failed to materialize, Shapland pivoted Tule into a profitable "lifestyle business" before CropX acquired it in January 2023, valuing Tule's canopy data as a complement to its own soil sensors rather than as a standalone product.[3][4]
The acquisition marked the end of Tule's independent run but not its technology's life: CropX folded Tule's FieldStat into its platform as the "ET Index," enhanced with a neural network, and all Tule employees joined the acquirer.[5]
Image 1 / 2
Tom Shapland's path to Tule began in the vineyards of UC Davis, where he earned a BS in Viticulture and Enology and a Doctorate in Horticulture and Agronomy. During his PhD work, he developed the underlying evapotranspiration (ET) measurement technology that would become Tule's core product — a method for measuring the actual water use of a farm field using a surface renewal energy balance approach.[2]
The founding story is unusual in that Shapland's initial attempts to build the company failed at every conventional step. He couldn't find a co-founder, investors, or employees. The breakthrough came not from a pitch deck or a prototype, but from farmer commitments. Shapland told the Way of Product podcast: "Traction unblocks everything. Sales unblocks everything. As soon as I got a bunch of farmers to agree to sign up to use it, even though I didn't have a product yet, they were like, yeah, build it and I'll pay for it. Then everything fell into place."[3]
Only after securing those farmer commitments was Shapland able to recruit Jeff LaBarge, a friend who had initially declined to work with him, to join as co-founder and CTO.[3] Tule entered Y Combinator's Summer 2014 batch with a team of eight, raising a seed round that sources place at either $120K or $600K, with investors including Omron Ventures and 500 Startups.[1][6][7]
Shapland's reflection on the early days reveals a founder who understood that his advantage was not experience but its absence. He told Way of Product: "The secret sauce I had is what every first-time founder has, and that is naivete. You just don't know how hard it's going to be. The challenge for my second time, for the second company I started, that naivete just isn't. Mostly, I look at the task and think, oh man, it's so hard and there's so much luck involved."[3]
Tule's product line evolved through three distinct phases, each building on the last but never escaping the hardware foundation.
The hardware sensor. Tule's core product was a physical sensor, developed at UC Davis, that measured Actual Evapotranspiration (ETa) — the actual water use of a farm field — over a 1 to 10 acre area. The company described it as "the only commercially available device for measuring Actual Evapotranspiration."[2] The technology used a surface renewal energy balance approach, which CTO Jeff LaBarge explained on Hacker News in 2017: "Tule Technologies (YC S14) is the commercial vendor of sensors that can measure evapotranspiration... the sensors measure ET through an energy balance approach called 'surface renewal.'"[8] For a grower, the sensor replaced the labor-intensive and error-prone practice of estimating water use from weather data or taking manual pressure chamber readings.
FieldStat. The software layer, FieldStat, translated raw ET data into a Water Stress Coefficient (Ks) — a measure of actual plant water use compared to potential water use. This metric correlated with leaf water potential (measured via pressure chamber) and was validated by 60 years of established science first introduced in the FAO 56 Evapotranspiration manual.[14] FieldStat won a Wine Business Monthly Innovation + Quality Award in 2018, signaling industry recognition.[9]
Tule Vision and Rover Vision. The AI phase began around early 2020 with Tule Vision, a computer vision app that allowed grape growers to take midday leaf water potential readings by recording short iPhone videos of their vines. Photos were fed into an AI model that reported readings in real-time.[10] The model was trained on "thousands of grapevine water stress measurements from Tule sensors and photos taken by field technicians."[15] Rover Vision extended this to counting grape clusters via smartphone camera, with a beta feature enabling users to mount an iPhone to an ATV and passively collect readings while driving down vineyard rows.[11] In August 2022, Suzuki Motor USA partnered with Tule to promote Tule Vision with KingQuad ATVs.[12]
The product trajectory is telling: Tule kept adding software and AI layers, but the hardware sensor remained the foundation. The AI products were trained on sensor data, meaning the hardware was not just a revenue stream but the data moat for everything else.
Tule's customers were high-value specialty crop growers: winemakers tending coastal California's most prized vineyards and agronomists farming thousands of acres of tree crops in California's inland valleys.[2] This was a deliberate niche choice. These growers faced acute water stress pain points — coastal vineyards where water stress directly affects wine quality, and inland tree nut orchards where water is both scarce and expensive. The customer base was concentrated enough that Tule could build deep relationships, but small enough that it capped the company's addressable market.
The precision irrigation market was real but fragmented. Tule's niche — high-value specialty crops in California — was a subset of an already specialized segment. The company never disclosed revenue or customer counts, and no public data exists on its market share. The absence of disclosed numbers is itself a signal: Tule's market was likely too small to support venture-scale returns, which is consistent with the company's inability to raise a next round.
Tule's competitive position was defined by a fundamental asymmetry: it measured water use from above ground (canopy), while most competitors measured from below ground (soil). CropX, the company that would eventually acquire Tule, was a soil sensor company. Other competitors included soil moisture probe vendors like Sentek and Decagon, and weather-based irrigation scheduling services like CropMetrics.
The structural problem was that Tule's above-ground approach was scientifically superior but commercially harder. Soil sensors are cheaper, easier to install, and require less interpretation. Canopy-based ET measurement required a PhD-level understanding of plant physiology to sell, and the hardware was more expensive to manufacture and maintain. Tule was competing on a dimension — measurement accuracy — where the marginal value to growers was real but hard to communicate, while competitors competed on dimensions — price, ease of installation — where the value was immediately obvious.
The competitive landscape also shifted as platforms moved. CropX was consolidating the precision irrigation space through acquisitions (Tule was its fourth since 2020), and its soil-to-sky strategy meant that Tule's canopy data was more valuable as a complement to soil sensors than as a standalone product.[16] This was the classic platform absorption pattern: a feature that incumbents could absorb more easily than a startup could scale independently.
Tule's revenue model was hardware sales plus software subscription. The sensor was a capital purchase, and FieldStat and Tule Vision were recurring software services. The company never disclosed revenue, pricing, or unit economics — a notable absence that suggests the numbers were not venture-scale.
The funding picture is murky. Crunchbase lists two rounds total: a "Venture - Series Unknown" round and a "Seed" round.[17] Sources conflict on the seed amount — $120K per startupintros.com, $600K per a former employee's LinkedIn profile.[6][7] The company never raised a Series A. With a team of eight at YC in 2014 and no disclosed later rounds, the total capital raised was likely under $2M — a fraction of what a hardware company typically needs to scale.
The unit economics were structurally challenging. A hardware sensor that measures over 1-10 acres means a grower with 1,000 acres needs 100-1,000 sensors. At even a few hundred dollars per sensor, the hardware cost per customer was significant, and the software subscription revenue per acre was likely modest. The company's pivot to a "profitable lifestyle business" — Shapland's own words — suggests the model worked at small scale but could not generate venture-scale returns.[3]
Tule's traction was real but niche. The company won two industry awards: FieldStat's Wine Business Monthly Innovation + Quality Award in 2018, and Tule Vision's Top-10 New Product recognition at the 2022 World Ag Expo.[9] The Suzuki partnership in August 2022 was a notable co-marketing win, pairing Tule Vision with KingQuad ATVs.[12]
The customer base — coastal winemakers and inland tree nut agronomists — was exactly the right niche for a scientifically rigorous product. These growers were sophisticated, water-stressed, and willing to pay for precision. But the niche was also the ceiling. No public data exists on customer counts, churn, or revenue growth, and the company's inability to raise a next round despite this traction suggests the numbers did not support a venture thesis.
Tule's fundamental problem was that it built a hardware company in a market that could not support hardware-scale capital requirements. Shapland was explicit about this: Tule was "initially on the venture path, but so many companies startups. We weren't able to raise the next round. We were fortunate enough to be in a position where we were profitable and we ended up pivoting the company into a more lifestyle, quote unquote, lifestyle business."[3]
The hardware trap has three compounding mechanisms. First, hardware requires upfront capital for manufacturing, inventory, and field deployment — capital that Tule never raised beyond seed. Second, hardware has a slower sales cycle than software, because growers must install, test, and validate the device before renewing. Third, hardware creates a data moat that is expensive to maintain: Tule's AI products were trained on sensor data, meaning the hardware was not just a product but the foundation of the entire data flywheel. When the hardware couldn't scale, the AI products couldn't either.
The attempted remedy was the pivot to software and AI. Tule Vision and Rover Vision were smart moves — they leveraged the sensor data into products that didn't require hardware installation. But the AI products were trained on sensor data, so they couldn't escape the hardware dependency. The Suzuki partnership was a creative distribution hack, but it was a co-marketing deal, not a scaling strategy.
The non-obvious structural mechanism in Tule's failure is the relationship between measurement granularity and market size. Tule's sensor measured ETa over 1-10 acres — a granularity that was perfect for high-value specialty crops but useless for row crops like corn or soybeans, where a single field can be hundreds of acres and the per-acre value is too low to justify sensor density. This granularity choice, driven by the science of surface renewal, locked Tule into a niche that was too small for venture-scale returns.
The counterfactual is instructive: if Tule had built a lower-cost, lower-accuracy product that could serve row crops, it would have competed directly with soil sensor vendors on price — a battle it would likely have lost. If it had built a higher-granularity product for even more specialized crops, the market would have been even smaller. The science was the strategy, and the science capped the market.
Tule's acquisition by CropX reveals the deeper structural dynamic. CropX CEO Tomer Tzach framed the acquisition as complementary: "With Tule's canopy data incorporated into the CropX system, CropX is adding a new and powerful dimension to the world's most complete precision irrigation solution."[4] The acquisition gave CropX entry into drip-irrigated specialty crops in California and created a "soil to sky" solution.[16]
This is the platform absorption pattern: Tule's canopy data was more valuable as a feature of CropX's soil-first platform than as a standalone product. The acquisition price was not disclosed, but the structure — all employees joining CropX, FieldStat renamed ET Index, Tule's technology enhanced with a neural network — suggests an acqui-hire-plus-technology deal rather than a strategic acquisition at a venture-scale multiple.[5]
The consensus framing of Tule's story is that it was a good company that couldn't raise a next round. The contrarian angle is that Tule was never a venture-scale business, and the YC path was a detour rather than a destination. The evidence: Tule raised only seed funding, never disclosed revenue, and pivoted to a lifestyle business before the acquisition. The company's own founder described the venture path as something Tule was "initially on" — implying it was a default rather than a conviction.[3]
The counter-counterargument is that Tule's technology was genuinely differentiated and its AI pivot was working — Tule Vision won a Top-10 New Product award at the 2022 World Ag Expo, and the Suzuki partnership showed commercial momentum.[9][12] But the awards and partnerships did not translate into a venture-scale growth curve, and the acquisition valued Tule as a complement, not a standalone business.
The precision irrigation category was structurally consolidating during Tule's lifetime. CropX's acquisition spree — four acquisitions since 2020 — was consolidating the fragmented market into a platform play.[16] A standalone hardware company in this category faced a shrinking window: either scale to platform status or become an acquisition target. Tule's science was strong enough to make it an attractive target but not strong enough to make it a platform. The outcome was determined less by Tule's execution than by the category's consolidation dynamics.
Hardware companies need hardware-scale capital. Tule raised seed funding and never closed a Series A. A hardware product requires manufacturing, inventory, and field deployment capital that software companies don't need. Tule's pivot to software and AI was the right instinct, but the AI products were trained on sensor data, so the hardware dependency was inescapable. If you build hardware, raise like a hardware company or don't build hardware.
The science was the strategy, and the science capped the market. Tule's surface renewal ET measurement was scientifically superior but locked the company into a 1-10 acre granularity that only made sense for high-value specialty crops. The niche was real — coastal winemakers and tree nut agronomists — but too small for venture-scale returns. Tule's technology was a feature of a larger platform, not a standalone business, and CropX's acquisition confirmed this by absorbing the canopy data into its soil-first system.
Traction unblocks everything, but only if the traction is venture-scale. Shapland's insight that "traction unblocks everything" was validated — farmer commitments unlocked his co-founder, investors, and employees.[3] But the traction that unblocked the founding was not the traction that unblocks a Series A. Tule's customer base was loyal and its awards were real, but the growth curve did not support a venture thesis. The lesson is not that traction doesn't matter — it's that the type of traction matters.
First-time founder naivete is a real asset, but it doesn't compound. Shapland's reflection that naivete was his "secret sauce" is honest and useful.[3] But naivete is a one-time resource. When Shapland started his second company, Canonical AI, the naivete was gone, and the company failed. The lesson is not to preserve naivete — that's impossible — but to build the systems and networks that replace it before you need them.