The mood at the World Agri‑Tech Innovation Summit in San Francisco was unmistakable: agentic AI has arrived, and the global agri-food sector is bracing for a consequential shift.
Across panel discussions, private roundtables and corridor conversations, anticipation and enthusiasm ran high, as speakers framed agentic AI as a transformative force capable of unlocking new frontiers of agricultural innovation and reshaping decision‑making across the value chain.
Beneath the excitement, however, was a clear recognition that agentic AI may shift power dynamics in agriculture – from how decisions are made to who makes them – prompting serious discussions about governance, data stewardship and responsible deployment.
SWARM Engineering: “Everyone wants to talk about AI”
Speaking to AgTechNavigator, Shail Khiyara, CEO of SWARM Engineering – a platform that uses advanced agentic AI to automate agricultural decision-making – said the industry’s AI conversation is still too narrow.
“Everyone wants to talk about AI,” he said, but noted there was not enough discussion around agentic intelligence and decision intelligence in ag – where the real transformation is beginning.
Conversations, he explained, fixated on data quality, data autonomy and data cleaning, rather than the next leap; how AI agents can turn that data into decisions and automated actions.
“Decision intelligence is not well understood,” he added. “People are still using Excel spreadsheets, making avoidable errors.”
From months to minutes: Agentic AI in action
To illustrate the impact of agentic automation, Khiyara pointed to Agrovision, the largest berry producer in Peru.
The company must coordinate the movement of 10,000 seasonal workers daily across a region vulnerable to natural disasters and logistical disruptions – an operational planning problem that previously took months.
“With AI agents and decision intelligence,” he said, “they are able to do that within minutes.”
Another example comes from Danper, Peru’s second‑largest fruit producer, which manages a complex matching process between fruit sizes, pallet configurations, shipping lanes and forward contracts – all previously done via spreadsheets by just two staff.
“That brings errors and volatility,” Khiyara said. “Shipping lanes change, pallet sizes aren’t accurate. So Danper came to us through Agrovision to build an AI‑driven system, with a human in the loop, to manage the whole cycle. That’s the kind of problem we are starting to solve in this space."
AI as a response to rising volatility and squeezed margins
Khiyara argued that the rapid rise in global agricultural volatility – from climate disruptions to freight uncertainty and geopolitical shocks – is driving a surge in demand for agentic AI.
“If you’re dealing with volatility, if you’re trying to do more with less, and if your margins are depleted like they are among fertiliser and grain companies, you need AI to solve that,” he said.
The technology, he argued, is increasingly seen not as a nice‑to‑have but as a core capability for safeguarding margins and enabling faster, more accurate responses to market and environmental uncertainty.
The three barriers holding AI back
Despite the excitement, Khiyara said three themes dominated conversations about AI adoption:
- Data quality – “Top of mind everywhere.”
- Trust and governance – organisations want transparency in how decisions are made.
- Customer creativity – or the lack of it.
“The ability to think about what AI can do for me, that’s still a challenge,” he said.Many agrifood firms rely heavily on experiential knowledge held by long‑tenured staff, but lack the structured data to codify it. “The more data you have,” he added, “the more value emerges in places you weren’t even looking.”
