No More Heroes (Any More)

Did you know that among the kiwifruit orchards in the Bay of Plenty this year will be beehives holding BeeHero sensors? What on the surface may seem like hive-auditing assistance, with the potential to gain beekeepers better pay for their services, could prove to be just the opposite – a company that plays the role of hive broker and removes beekeeping expertise entirely, leaving existing hive owners further impoverished. We examine what we know about how hive-management technologies such as BeeHero work, who is behind them, and how they hope to recoup the nine-figure sums invested in them thus far. It’s a story of money and power, that has little to do with beekeepers and their needs.

By Dave Black

Not too long ago no-one cared very much about a few beekeepers and their hives, but now it seems they are deemed essential to the world’s food security, and for the livelihoods and health of millions. If only that were true when it comes to funding apicultural science or settling the season’s invoices. The response to this dependence has varied.

Naturally, one answer is to try and cut honey bees out of the loop. Melibio, acquired in May by FoodYoung Labs in Switzerland, and Israel’s Bee-io, will make ‘honey’ in a vat and save the “overused and abused” bees. BloomX (formerly BumblebeeAI) are already building mechanical pollinators to replace bumble bees as pollinators – they’re too fussy and “not optimal” – and BeeWise hopes it can replace beekeepers with ‘active beekeeping’ (software and robots) that can do a better job[i], and is prepared to bet almost NZ$300 million (of other people’s money) that they can.

 Beewise hopes to replace beekeepers with robotic hives such as this, and is prepared to bet almost NZ$300 million (of other people’s money) that they can.

It’s tempting to think beekeeping just exists in its own little bubble, but there are big societal forces at work to make sure that isn’t so. The sheer scale of honey bee colony losses in recent decades, particularly in the US, has certainly fuelled a renewed interest in remote hive monitoring and analysing the information being collected. Collecting information about hives is nothing new of course, but the scope for using and sharing that information is much, much greater when it is digital data. The measuring instruments we now use have assumed the form of digital sensors; even vision and sound is now encoded digitally.

Opportunity Knocks

In a sense, that’s the easy bit, and always was. The difficult bit is to make sense of the sensors sensing, and that’s still true too. The promise large ‘Tech’ corporations are making now is that, if they have enough data (your data), using tools with impressive names like ‘Machine Learning’ (ML), Deep-Learning, and Artificial (and ‘Convoluted’) Neural Networks (ANNs/CNNs) – which in the current hype-cycle we wrap up with a bundle of other things and call ‘Artificial Intelligence’ (AI) – they can make correlation equal causation. With enough data about the goings-on within a hive they (or we) will glean complicated patterns that will actually predict what the hive will do from what it is doing now. No analysis required.

Being able to ‘save the bees’, using ‘AI’, ticks just about every box when it comes to the metaphorical ‘Dragon’s Den’, and sure enough, investment funding poured in. What investor isn’t going to back an ‘entrepreneur’ doing that?  Remote hive monitoring has morphed into something quite different – precision agriculture is looking at precision pollination. BeeHero, another company started in Israel, is a case in point and, for us down in New Zealand, a timely example as Zespri will be trialling the product with some growers and beekeepers this spring.

Israel is a case study for economists trying to understand how to promote entrepreneurial activity and capital growth in advanced economies[ii]. It is a story about ingenuity, but it’s also a story about money and power, things beekeepers have never had. What really distinguishes these companies is the eye-watering amount of cash they have to spend in comparison with the other players in this sector.

BeeHero has so far raised more than NZ$100 million. It speaks volumes that they are marketing their technology for intensive almond production, and to the agro-chemical corporations whose products contribute to the degradation of soil health and insect biodiversity. Despite what they say, these companies are not using smart hives and precision pollination as a way of addressing the health of pollinators, CCD, or honey bee welfare. By targeting agribusiness, like the robotic beehives of BeeWise, BeeHero is following the money. Selling stuff to beekeepers doesn’t credibly sustain market valuations approaching hundreds of millions of dollars.

 BeeHero data collectors will be placed in beehives in the Bay of Plenty during kiwifruit pollination this spring, but ownership of the data collected is not likely to pass to either beekeepers, or the growers footing the bill.

A Digital ‘Black-Box’

BeeHero’s ‘mission’ is to “drive value by tracking and monitoring bee activity in crops to deliver quality data for quantifiable pollination.” It’s not an original idea, but they have at their disposal the determination, expertise, and financial muscle to actually try and execute it. Apart from giving growers a form of visibility and accountability for the pollination units they hire, BeeHero’s data collection tries to compare the colony’s ‘metrics’ with activity and outcomes in the crop being grown. In their words, “Learning models predict colony issues and measures bee activity and pollination efficiency and transforms it into valuable, actionable insights.”

The system starts with sensors in the hive that capture temperature, humidity, light levels, location, hive orientation, and sound, all in a single, robust, bee-proof unit attached to a brood frame. This unit connects to an adjacent external ‘gateway’ using Bluetooth, and the gateway transmits the data to ‘cloud’ hosted servers using a GPRS modem. There are also sensors in the target crop ‘listening’ for the sound of foraging bees, filtering out background noise like wind, traffic, and other farm equipment noise, and sending that to a local gateway by Bluetooth.

 BeeHero also uses field-based data collectors to monitor bee activity, as seen here in an onion crop.

The data is stored, classified into types, and analysed using several mathematical and statistical strategies (‘algorithms’), and compared to other datasets and computer models to try and make sense of it – the proverbial ‘black-box’ AI. The output is selected, rendered, and reported in graphical displays designed to inform particular audiences, like a beekeeper using a mobile phone, or a grower using a PC or laptop.

These displays can be integrated with other tools and data. For example, for growers BeeHero uses Google’s Maps to display orchard geography, and includes a calculator application that will calculate hive stocking rate based on what it knows about that geography. It can also display a ‘heat-map’ of bee ‘activity’ to highlight areas that might not be getting the attention they deserve.

Silicon Wadi

BeeHero was born in a place now known colloquially as ‘Silicon Wadi’, Israel’s analogue of California’s Silicon Valley, a global hub for tech innovation comprised of a cluster of companies specialising in software, communications, electronics, hardware design, and internet technologies. The phenomenon, besides being an academic curiosity, has been reported in more sensational terms, like by the notable Stacy Perlman, an investigative journalist with the New York Times magazine[iii].

The consensus is that the factors which contribute most to Israel's competitive advantage were immigration and its compulsory military service, but there are undoubtedly others. Israel also made education a national priority, especially science and technology, and achieved very high English literacy. It managed to develop an advanced network of cheap and reliable telecoms providers and the government offered a number of financial incentives to firms to locate high-tech research activities locally, so that a steady stream of US firms found Israel a prosperous place to invest. These companies saw the benefit of focusing on areas that were information-based, and in areas of expertise that were relatively close to military interests like electronics and wireless communications.

 Known as ‘Silicon Wadi’ (a play on California’s Silicon Valley), in Israel the high-tech industry is flourishing and numerous beekeeping companies have benefitted. Whether they can return a profit to investors is still unproven.

The country’s technological prowess is closely linked to the activities of its military forces, including the development of defence and surveillance technologies. Several key military units take the brightest high school graduates, trains them in intensive science and engineering programs, and disperses them into military roles. Israel's universities are closely allied and young innovators know each other through their military service, their most important networking and workforce training opportunity.

BeeHero fits this mould perfectly. The company was founded by four university pals. Omer Davidi, had a military background in cybersecurity and machine learning[iv]. Yuval Regev served seven years in an elite Israeli intelligence unit[v]. Itai Kanot’s happened to have a father who was chairman of the Israel Beekeepers Association, and the only woman, Michal Roizman, had a background in robotics and interactive media but left for another start-up in2020. [vi]

Who’s Paying the Piper?

BeeHero relies on the data to provide accurate metrics on the strength and health of each colony as an effective pollinating force, apparently accurate enough that it can compare its measurement to a per-frame payment, rather than a per-hive payment. That way, beekeepers delivering strong hives should earn higher rates than beekeepers delivering weaker hives.

They’re not charging for the hardware, but they are getting your data for free.

Growers using BeeHero should pay costs competitive with conventional pollination contractors, but the data is supposed to earn a benefit in higher yields and better ‘value for money’. Growers can pay extra for data queries or AI ‘insights’ too. When questioned last year by Australian Rural & Regional News BeeHero did not respond to questions about who owns the data or to whom it would be made available[vii].

It’s worth pointing out that to date there are no profitable AI companies[viii]. Even though they get data at no cost, really all their products are supplied as ‘loss-leaders’, there’s revenue but no profit. It’s worth wondering what will happen as users of these tools are confronted with paying the real costs, which must happen if the sector is to have a sustainable future.

Not a Tool, a Belief System

Fundamentally, BeeHero are hive brokers, middlemen negotiating a service supply and clipping the ticket. Their point of difference from the competition is information, they claim that they have more ‘knowledge’[ix] about the crop and its pollinators and that they use it to benefit both parties to the trade. BeeHero also has the opportunity to use that information, which they hold exclusively, for their own benefit.

They are not making the invisible, visible. Growers still do not really know if they are paying for good pollination hives, but they believe BeeHero does. Apiarists may think they know, at least what the hives should be like, but they too have to accept what BeeHero tell them. It’s still all about faith and trust; the rest is just smoke and mirrors.

It’s a sector where glossy marketing websites triumph in a lesson of style over substance. I have not found any independent evidence that the data provides good information; all the models used for the assessment are proprietary and not available for researchers to view. I imagine BeeHero’s users will be tied up by Non-Disclosure Agreements.

Epistemologically, AI tools are deeply problematic[x]. It’s not just me. Washington State University’s Brandon Hopkins “has not seen researchers independently validate the quality improvement the ag tech companies tout”. A recent paper on bioacoustic monitoring notes “Imperfect detection pervades all sampling methods”[xi] although, as a positive note, no alternate insect surveying method was any better. With a more prosaic comment on the technology, talking about the recent (2025) losses, president of the California State Beekeepers Association, Ryan Burris, pointed out, “These technology companies [Beewise/BeeHero] didn’t see it coming... They didn’t warn us about it. They didn’t stop it.”[xii]

A hard-working honey bee searches for kiwifruit pollen.

Insider Perspectives

The beekeepers I meet often chat about the problems they face: running bees, at scale, is physically demanding; any form of plant or animal husbandry is risky, and then there is financing your loss-making hobby. They talk about challenges with all sorts of things – managing people, quality, safety, records, compliance, money, logistics, machinery, traceability, dishonesty, pollution, competitors, uncertainty, change, and acts of God. While it’s nice to know more, and know more detail, being clueless about the status of their colonies is not something they need help with. For people who aren’t beekeepers, it must be said, beehives are their own kind of mysterious ‘black-box’.

There have been many designs for digital hive monitors in use and, although in general terms beekeepers are not unaware of the potential benefits, the uptake has been pretty limited. Certainly, some had difficulty with the ‘prototype’ nature of the equipment in the beginning and some were sceptical that reliable conclusions could be taken from the data (the jury is still out). Others were critical of the principle itself, uncomfortable with providing information to an unreliable third-party who could use it to gain an unfair advantage, or pass it (or lose it) to someone who would. They were wary of being spied on. There is a feeling about a ‘solution’ looking for a ‘problem’, and troubling sense that focusing on data, or apiculture’s imperfection, is not tackling the root cause of a problem that modern ‘industrial’ agriculture itself contributes to, and which it employs honey bees to remedy.

Standards and Measurements

Beekeepers get all excited about the idea we might get paid for quality. In simple terms if they provide colonies that exceed the agreed standard (more frames of bees) they earn a premium. It’s curious that we are confident that a company paid by the grower is interested in increasing the cost to the grower and will be a fair arbiter.

If the existing system works as intended the pollination units provided to the standard should pollinate all the flowers there are – that’s the purpose of a ‘standard’. It’s not possible to ‘do better’ than full pollination, so the benefit of exceeding the standard must be marginal or nil, and it’s certainly a fact that over-pollination is detrimental (so is over-cropping). It might be galling, but to me the scope for quality premiums is limited, and I doubt the wisdom of setting up BeeHero to be in charge of the ‘standard’.

We must also know how quality is to be measured. If it’s more frames of bees for example, how are they counting frames of bees? We know counting bees is hard, and that all methods, short of euthanising the bees and counting one by one, are inaccurate and only estimates. How inaccurate? We need to do some work to determine that, but regardless BeeHero doesn’t count bees either – it too estimates. AI would have to be ‘taught’ to count, and it's being taught with our inaccurate estimates! Its estimate must be faster, and more complete; it can include every hive rather than a sample, but scaling up a bad estimate won’t improve it.

The current system of beehive standards for kiwifruit pollination has been set with the potential to achieve 100% pollination success for orchards, so is there any added value to be gained for beekeepers by exceeding standards?

Irrelevant or Irreplaceable?

The relative freedom of beekeepers within commercial agriculture depends on their ownership of bees and their hives, and their local, inter-generational, corporeal knowledge. In a ‘precision’ pollination system what were simple wooden boxes become, or are augmented by, proprietary digital technologies owned or licensed by multi-national tech companies. The property, knowledge, and labour of the beekeeper, and their bees, becomes irrelevant and fungible[xiii]. Be careful what you wish for.

After all, it's not called Artificial Wisdom.

References

[i]Beewise Impact Report 2024

[ii]De Fontenay, C., Carmel, E., 2004. Israel’s Silicon Wadi: The Forces behind Cluster Formation, in: Bresnahan, T., Gambardella, A. (Eds.), Building High-Tech Clusters. Cambridge University Press, pp. 40–77. https://doi.org/10.1017/CBO9780511802911.005

[iii]Perman, Stacy, (2005) Spies, Inc., business innovation from Israel's masters of espionage, p241. Pearson/Prentice Hall, ISBN10 0-131-42023-2

[iv]https://qz.com/1633976/israeli-startups-are-building-artificial-bees-to-support-pollination

[v]https://www.startup.review/company/beehero-io-machine-learning-algorithms-and-low-cost-sensors-to-optimize-pollination

[vi]https://forbes.co.il/e/queen-bee/

[vii]https://arr.news/2024/08/05/beehero-launches-pollination-insight-platform-2-0-for-data-driven-pollination-predictions-and-improved-in-field-monitoring-across-the-globe/

[viii]For a rather lengthy 'tech industry' analysis see Edward Zitron, The Hater's Guide To The AI Bubble, 21-07-2025. https://www.wheresyoured.at/the-haters-gui/

[ix]In a strictly academic sense, AI can not produce knowledge (justified, true, belief), but it can produce information

[x]A critique is beyond the scope here, but see https://helenbeetham.substack.com/p/superintelligent-research

[xi]Hearon, L.E., Johnson, L.H.P., Underwood, J., Lin, C.-H., Johnson, R.M., 2025. buzzdetect: an open-source deep learning tool for automated bioacoustic pollinator monitoring

[xii]https://goodfruit.com/bee-businesses-abuzz-with-high-tech-beehives/ Ryan Burris (California State Beekeepers Association), who owns Park Legacy Queens of Palo Cedro

[xiii]Nimmo, R., 2025. Digital hives, nonhuman work and the real subsumption of nature: Fixing pollination in capitalist agriculture. Environment and Planning E: Nature and Space 8, 1112–1131. https://doi.org/10.1177/25148486251336609