Halter's Clank AI cuts engineering time in farm tech

Halter has introduced an AI agent called Clank to automate engineering work across its connected livestock platform, saving more than 215 hours of engineering time.

Built on Amazon Bedrock using Anthropic Claude, Clank investigates technical issues, prepares code changes and handles routine workflows for the New Zealand-founded agritech group.

Over an 11-week period, the agent automated more than 90 weekly tasks and processed more than 2,500 code changes. Halter said an AI-generated code change can now be reviewed and embedded into its systems in 48 minutes.

The move highlights how agricultural technology companies are using generative AI in internal operations, not just customer-facing products. At Halter, the software supports a platform that lets farmers manage cattle remotely through connected collars and a mobile app.

The system allows farmers to set virtual boundaries for herds and monitor animal wellbeing and behaviour. That requires Halter's engineering teams to keep a large, always-on technology platform running across multiple countries and time zones.

Halter says the business has more than 16,000 Amazon CloudWatch alerts monitoring components across its systems. Before Clank, engineers responded to more than 40 alert tickets a day when technical support was needed, manually investigating root causes and deciding how to respond.

Clank now analyses alerts automatically, runs root-cause investigations and prepares fixes for engineers to review. It is integrated with Slack, GitHub and Linear, allowing it to work directly within the company's existing workflows.

When an alert is triggered, the system retrieves logs and other data before sending the issue to Claude through Amazon Bedrock for analysis. It can then generate a code change before an engineer starts work, shifting the team's role from manual triage to review and approval.

Halter described the development as part of a broader effort to reduce time spent on administration and operations, both for farmers and internal teams. Its products are designed to let livestock farmers manage herds remotely, including moving animals without being physically present on the farm.

"With Halter, farmers no longer need to be out on the farm at 4am to move animals. They can manage their operations remotely and spend more time focusing on their business and their families," said Kurt McAlpine, cloud architect, Halter.

That customer proposition depends on system reliability, especially because the company's technology is tied to day-to-day livestock management. Halter said this shaped its approach to using AI in engineering operations.

"There's an animal on the other end of that technology, so it's very important to us that everything is robust and scalable," said McAlpine. "We chose Amazon Bedrock as it gives us the reliability and data privacy we need to be secure and confident in deploying Clank."

AI in farming

The announcement comes as New Zealand agricultural businesses consider a larger role for AI. Research cited by AWS found that 61% of agricultural businesses in the country expect AI to transform large parts, or all, of their industry within five years, while only 9% said they felt very or fully ready to adopt next-generation AI technologies.

For Halter, the immediate use case is operational efficiency inside the business rather than direct automation of farm decisions. By using AI to handle repetitive engineering work and the first stage of incident response, the company is seeking to reduce downtime risk and free staff for product development.

One internal measure points to the scale of that shift. "On an average day, Clank works the equivalent of 231 hours," said Antony Southworth, lead data engineer, Halter. "That's nearly 29 times a typical engineer's workday - at an inference cost of approximately US$30 per hour, making AI-driven incident response both faster and significantly more cost-effective than manual triage."

Halter said the time saved by Clank is being redirected into work on new models and improvements to the farmer experience. The company argued that reducing engineering effort on platform incidents creates more scope to develop features that help farmers manage herds with fewer interruptions.

McAlpine linked that internal shift to the wider adoption of AI across agriculture. "AI is completely reshaping what's possible for our industry, and we at Halter are proud to support farmers through this next phase of evolution to make their lives easier than ever," he said. "AWS have helped us to transform into an AI-led business, skyrocketing our opportunities to scale our farming technology to the world and be a leading key agricultural exporter for New Zealand."

AWS said the Halter deployment reflects growing interest in practical AI use cases among New Zealand businesses. "Across New Zealand, we're seeing organisations embrace AI to solve real-world challenges, improve productivity, and create new opportunities for growth," said Manuel Bohnet, country manager, New Zealand, AWS. "Companies like Halter demonstrate how cloud and generative AI can help agribusinesses innovate faster, operate more efficiently, and build leading, globally competitive businesses - right from the heart of New Zealand."