AI in Agriculture in Aotearoa New Zealand: A Living Whitepaper

Updated 13 July 2026

Introduction

AI in New Zealand agriculture is now best understood as a set of production-grade tools tied to specific farm and orchard decisions rather than a broad, uniform “AI wave.” The commercial centre of gravity remains dairy, especially in livestock control, pasture intelligence, and farmer-facing decision support. Horticulture and arable are advancing through narrower but high-value use cases such as fruit-quality prediction, disease-risk modelling, and seed screening. This is happening in a sector of major economic weight: MPI’s June 2026 SOPI expects food and fibre export revenue to reach NZ$64.3 billion in the year to 30 June 2026, including NZ$28.6 billion from dairy, with growth also supported by stronger apple and kiwifruit exports. (mpi.govt.nz)

Executive Snapshot

• Dairy remains the most advanced AI segment. Halter’s virtual fencing and animal-management platform, Aimer’s pasture-intelligence system, and DairyNZ’s DAiSY assistant are still the clearest examples of AI being used in day-to-day farm operations. (halterhq.com)
• The biggest shift since 10 June 2026 is stronger institutional embedding. AI is moving from startup deployment into co-funded programmes, farmer support channels, and sector-wide delivery mechanisms, notably through DairyNZ’s new Responsible Dairy programme and Halter’s inclusion in Fonterra’s 2026/27 On-farm Solutions incentive scheme. (dairynz.co.nz)
• Horticulture AI remains narrower than dairy, but often strategically higher stakes. Recent momentum is concentrated around fruit-quality visibility before packout, early disease detection, and contamination screening, with Hectre, Lincoln Agritech/STELLA, and Plant & Food Research’s Hyperseeds work standing out. (auckland.ac.nz)
• Public-good AI is hardening into sector infrastructure. The On-Farm Emissions Calculator, AgResearch’s Map and Zap®, and MPI’s use of AI-enabled cameras in hornet eradication all show AI moving beyond private software into public research, biosecurity, and sector capability. (agresearch.co.nz)
• The binding constraints are still trust, integration, and proof of value. DairyNZ continues to frame GenAI as a support tool rather than a substitute for farmer judgement, and MBIE’s AI strategy material still highlights a national SME adoption gap. (dairynz.co.nz)

What Changed Since the Previous Update on 10 June 2026

• Aimer launched a new pasture stack on 30 June 2026 that combines proprietary satellite data, AI-powered smartphone measurement, and live paddock modelling. Aimer says it is already used on more than 650 farms with more than 10,000 pasture measurements a week. (aimer-farming.com)
• DairyNZ’s Responsible Dairy programme became the clearest new institutional signal. The seven-year programme is funded at NZ$45.85 million, with NZ$18.34 million from MPI’s Primary Sector Growth Fund, and will test next-generation systems and technologies across 35 to 40 partner farms. (dairynz.co.nz)
• Halter entered a more mainstream adoption channel via Fonterra. From 8 July 2026, eligible Fonterra farmers can apply for up to NZ$2,000 + GST toward Halter under the co-op’s 2026/27 On-farm Solutions initiative. (halterhq.com)
• Fieldays 2026 strengthened the “innovation funnel” story. Organisers reported a marked increase in AI-based solutions, and the Prototype Award went to Scanabull, whose WeighApp uses 3D LiDAR and AI to estimate cattle liveweights from a phone scan. (fieldays.co.nz)
• The national AI research-platform outcome still does not appear to be publicly finalised. On MBIE’s AI Research Platform page, the phase-two timetable is visible and MBIE says “an update on a timeline of announcements will be made in due course”; no final platform award is shown on the public material visible in July 2026. This is an inference from the current public pages, not a formal delay notice. (mbie.govt.nz)

Current News and Market Developments

1) Dairy and livestock AI are moving deeper into the operating layer

Halter remains the clearest scaled example of AI in New Zealand agriculture. Its 28 April 2026 direct-to-satellite launch matters because it reduces a core deployment bottleneck: connectivity. Halter says the new One NZ Satellite/Starlink integration removes the need for on-farm infrastructure for beef operations, could expand access for New Zealand beef farms by at least 20%, and accompanies new tools for reproduction, animal behaviour, and precision pasture management. Halter also says it now serves more than 2,000 farmers and ranchers across New Zealand, Australia, and the US, and has sold one million collars. (halterhq.com)

The more important July development is that Halter is no longer only a high-growth agritech product; it is also being channelled through mainstream dairy incentives. Its inclusion in Fonterra’s 2026/27 On-farm Solutions scheme suggests wearable AI is moving closer to accepted farm-improvement spend, especially where it can be framed around efficiency and emissions outcomes. That does not prove universal adoption, but it is a strong signal that AI is entering ordinary farm capital decisions rather than sitting only with early adopters. (halterhq.com)

Aimer is the other major livestock-side signal. Its 30 June 2026 launch of AIMER Satellite pushes pasture AI from smartphone measurement into a blended system of satellite observation, farm-specific paddock modelling, and AI-assisted estimation. The company says direct smartphone measurement exceeds 90% accuracy, while the broader satellite-linked system is designed to maintain around 80% daily accuracy between observations. Those are company-reported figures, but strategically the launch matters because it addresses the central New Zealand dairy problem: pasture decisions have to be made daily, not just when measurements are convenient. (aimer-farming.com)

DairyNZ’s recent commentary reinforces that pasture-based fit is decisive. In its 16 June 2026 article, DairyNZ argued that New Zealand’s grazing model should shape how technologies are developed and adopted, and highlighted pasture productivity measurement as a high-potential area only if tools are easy to use, cost-effective, and relevant to day-to-day decisions. That framing aligns closely with the commercial logic behind Aimer, Halter, and other workflow-native systems. (dairynz.co.nz)

2) Sector institutions are becoming AI deployers, not just observers

DairyNZ is now using AI directly in farmer-facing delivery through DAiSY, its website assistant. DairyNZ says DAiSY draws from more than 1,100 pages and 880 tools and resources, and responds only from DairyNZ website content. That matters because institution-led AI lowers one of agriculture’s biggest barriers to adoption: trust. In practice, this is a retrieval-and-summarisation layer grounded in known sector material rather than an open-ended consumer chatbot. (dairynz.co.nz)

The new Responsible Dairy programme is an even bigger sign of institutional embedding. DairyNZ says the programme will bring together science, farming, finance, fertiliser, and technology partners — including Halter and Gallagher — to test stacked technologies, model next-generation systems, and accelerate adoption on commercial farms. The design implies that the next stage of AI adoption will be less about isolated tools and more about combinations of wearables, data systems, environmental measurement, and farm-system redesign. (dairynz.co.nz)

3) Horticulture and arable AI are still narrower, but the economics are sharp

Hectre remains one of the strongest commercial signals in horticulture. The Auckland-founded company raised an oversubscribed NZ$12 million Series A in February 2026 to expand its AI and computer-vision systems for fruit sizing, colour, and quality before produce reaches the packhouse. University of Auckland reporting says the raise will support expanded R&D and hiring. The strategic importance is straightforward: pre-packhouse visibility improves storage, grading, and sales decisions before value is lost. (auckland.ac.nz)

Lincoln Agritech’s STELLA work shows the export-risk side of horticulture AI. In Hawke’s Bay, the project is combining automated spore samplers, UAV and satellite imagery, and environmental monitoring to feed AI-powered risk models for bull’s-eye rot in apples. Lincoln Agritech links the work directly to New Zealand’s apple-export exposure, noting that apple exports reached a record NZ$1 billion in 2025. This is a strong example of AI being justified not by labour saving alone, but by market access, reputation, and storage-loss prevention. (lincolnagritech.co.nz)

In arable and seed systems, Plant & Food Research’s Hyperseeds project with the Foundation for Arable Research is one of the more concrete 2026 examples. The collaboration uses hyperspectral imaging and AI to pre-screen seed for contaminants, targeting a labour-intensive quality-assurance bottleneck. The importance here is not scale today, but breadth: AI is spreading into New Zealand agriculture where visual inspection, purity, and quality assurance have high economic value. (plantandfood.com)

4) Public-good AI is becoming more visible as sector infrastructure

The On-Farm Emissions Calculator, launched in October 2025, remains one of the most important enabling tools because it is the first calculator to apply the government-mandated standard released in December 2024. AgResearch says the platform is designed not only for direct farmer use but also so agri-tech providers and processors can embed the same methodology in their own systems later. In other words, it is less an end-state app than a standardised substrate for future advisory and optimisation tools. (agresearch.co.nz)

AgResearch’s Map and Zap® is a similar infrastructure signal on the weed-control side. The system uses AI to identify early-stage weeds and then map or laser-zap them, with AgResearch estimating weeds cost New Zealand agriculture and forestry at least NZ$1.7 billion annually. The significance is that AI is being developed for physical intervention, not just analytics. (agresearch.co.nz)

MPI’s yellow-legged hornet response adds a biosecurity example. In March 2026, Biosecurity New Zealand introduced AI-enabled cameras from the University of Exeter’s Vespa AI team to help monitor hornet activity and locate nests. That is not mainstream farm software, but it is still agricultural AI: it supports the protection of production systems through earlier and better surveillance. (mpi.govt.nz)

Research Overview

DairyNZ’s current AI research framing remains cautious. Its farmer-facing material says GenAI uptake is growing, but trust and accuracy remain key, and its formal guidance still positions AI primarily as a decision-support tool rather than an autonomous farm manager. (dairynz.co.nz)

The 2025 International Precision Dairy Farming Conference proceedings remain one of the best practical research windows into New Zealand conditions. A Southland chatbot case study reported that farm managers routinely monitor 4 to 7 applications across weather, pasture, soil moisture, and livestock data, and that farmer interest centred on synthesising fragmented systems rather than adding yet another standalone tool. (dairynz.co.nz)

AgResearch continues to represent the deepest public research bench in the field. It says it has more than 50 AI-focused projects in planning or implementation, including CT-scan analysis for livestock traits linked to methane emissions, feed efficiency, welfare, and meat quality. The AgResearch/Bioeconomy Science Institute portfolio suggests that AI in New Zealand agriculture is increasingly as much a science-platform capability as a startup category. (agresearch.co.nz)

At the science-system level, government settings are becoming more commercialisation-oriented. MBIE’s Science Investment Plan 2026–2036 says NZ$142 million has been invested through the New Zealand Institute for Advanced Technology in high-potential areas such as AI and quantum, while new IP rules took effect from 1 July 2026 to give publicly funded researchers more control over commercial outcomes. For agriculture, that matters because the next wave of value is likely to come from translation and deployment, not just research output. (mbie.govt.nz)

Case Studies

Halter: AI as farm infrastructure

Halter now spans virtual fencing, remote shifting, behaviour monitoring, reproduction tools, and pasture management, with direct-to-satellite capability extending use into remote beef systems. Its new place inside a Fonterra incentive scheme is a strong signal that the product is moving from standout innovation toward more standardised farm adoption channels. (halterhq.com)

Aimer Farming: pasture AI built for pasture systems

Aimer’s 2026 progress shows what a New Zealand-specific AI stack looks like: smartphone-first measurement, digital-twin paddock models, satellite augmentation, and direct alignment with grazing decisions. Its relevance comes from system fit more than headline scale. (aimer-farming.com)

Hectre: AI before the packhouse

Hectre’s raise confirms investor confidence in horticultural AI that helps growers and packhouses make earlier decisions on storage, grading, and sales. It remains one of the clearest export-oriented AI growth stories in New Zealand agriculture. (auckland.ac.nz)

Lincoln Agritech / STELLA: AI for export-risk management

STELLA is a high-value example of AI being used to detect what conventional sensing cannot: disease risk before symptoms appear in storage or overseas markets. In export horticulture, that kind of foresight is strategically important. (lincolnagritech.co.nz)

Scanabull: the widening innovation funnel

Scanabull’s 2026 Fieldays Prototype Award win shows how AI is spreading into smaller, narrower use cases with obvious farm utility — in this case, cattle liveweight estimation from a phone scan. That is exactly the kind of tool that can diffuse faster than generic “AI platforms.” (fieldays.co.nz)

Core Trends

1) Workflow-embedded AI is still outperforming general-purpose AI

The strongest systems are attached to an existing farm job: move cattle, estimate pasture, detect disease risk, screen seed, calculate emissions, or identify weeds. The more a tool produces an action rather than a generic insight, the better its adoption prospects appear. (halterhq.com)

2) New Zealand’s pasture-based dairy model continues to shape the market

DairyNZ’s recent material, MBIE’s AI strategy examples, and Aimer’s product design all point in the same direction: the highest-value tools in New Zealand are those built for grazing systems rather than imported from housed or row-crop contexts. (dairynz.co.nz)

3) Institutional channels are reducing adoption friction

DairyNZ’s DAiSY, the Responsible Dairy programme, and Fonterra-linked incentives suggest a maturing market in which sector bodies are helping validate, fund, and normalise AI use. That is an important shift from pure startup-led diffusion. (dairynz.co.nz)

4) Horticulture AI is often easier to justify where export risk is concentrated

Compared with dairy, horticulture AI adoption is narrower, but the economic case can be sharper because a single grading, storage, or disease-detection failure can destroy significant value. Hectre and STELLA fit that pattern closely. (auckland.ac.nz)

5) Public-interest AI is becoming part of the operating environment

From emissions accounting to biosecurity surveillance and non-chemical weed control, AI is no longer only a private farm-software story. It is increasingly part of New Zealand’s agricultural capability stack. (agresearch.co.nz)

Constraints and Risks

• Trust and validation remain decisive. DairyNZ’s guidance continues to emphasise trust, data quality, and the role of AI as support rather than replacement. (dairynz.co.nz)
• Interoperability is still a practical bottleneck. Research from the precision-dairy conference shows farmers already managing multiple apps and wanting better synthesis across systems. (dairynz.co.nz)
• Many of the strongest performance claims are still company-reported. That is especially true for commercial pasture, livestock, and wearable systems, so the sector still needs more independent, longitudinal validation. This is an inference based on the current mix of company announcements and sector commentary. (aimer-farming.com)
• The SME adoption gap still matters. MBIE’s AI state-of-play material says 68% of SMEs had no plans to evaluate or invest in AI, a material issue in a primary-sector ecosystem dominated by smaller operators and suppliers. (mbie.govt.nz)
• National AI research coordination is still in formation. Agriculture-relevant concepts are prominent in the AI Research Platform process, but the final public outcome had not been posted on the current MBIE material visible in July 2026. (mbie.govt.nz)

Conclusion

As of 13 July 2026, AI adoption in New Zealand agriculture is best described as practical, selective, and increasingly embedded in sector institutions. Dairy remains the most mature segment, led by livestock wearables, pasture intelligence, and trusted advisory interfaces. Horticulture and arable continue to adopt AI where quality, disease, and export assurance create clear economic leverage. Public-good AI is also becoming more consequential through emissions, weeds, and biosecurity. (halterhq.com)

The key change since the previous update on 10 June 2026 is not a single breakthrough model. It is the strengthening of the adoption layer: new pasture-intelligence capability from Aimer, new institutional funding and testbeds through Responsible Dairy, new farmer incentive pathways via Fonterra-linked support for Halter, and stronger evidence from Fieldays that AI is broadening into a larger pool of practical farm tools. (aimer-farming.com)

The near-term winners are still the same kinds of systems that have been leading for the past year, but the evidence base is stronger now. In New Zealand agriculture, the tools most likely to move from early-adopter enthusiasm into mainstream use are those that combine workflow fit, trusted grounding, low-friction deployment, and visible farm economics. (dairynz.co.nz)