.png){kind=small}
From Lemons to Lifelines: What Supply Chains Teach Us About AI Readiness
0
1
0
The Hidden Prerequisite for Agentic AI
Everywhere you turn, “agentic AI” is the new headline, systems that can reason, plan, and act with autonomy. But no AI system, no matter how clever, can operate intelligently if its data isn’t amenable: accessible, aligned, and actionable.
Data amenability is the true currency of readiness. It determines whether AI can learn, adapt, and respond to context or stall in confusion.
Two contrasting industries: healthcare and agriculture, reveal what happens when data is powerful but uncooperative. I spoke with Dhaval Desai, healthcare interoperability consultant and former Chief Architect at Macy’s, and Lakshmi Duvulapalli, Enterprise Data Architect at Limoneira. Different industries, same problem: fragmented ecosystems and legacy ontologies that block AI from scaling.
Lifelines: When Healthcare’s Data Stands Still
Healthcare sits at the intersection of technology and trust. It has oceans of data yet deep gaps in access.
“The U.S. healthcare system is very fragmented… the question is how we can logically connect those systems together and make a meaningful application using AI.” — Dhaval Desai
#Real Challenge: Fragmentation Has a Cost
Hospitals still run on EHRs built in 2004. HIPAA and TEFCA ensure security, but slow innovation. Providers fear that even a minor API change could expose patient data, turning six-month projects into year-long odysseys.
AI in healthcare thus remains technically powerful but procedurally constrained. Every innovation must navigate audits, validations, and workflow approvals before it can act.
Learning: Healthcare has data quality, but not data agility.
Even with interoperability tools like QHIN networks or “Agent DKI” summarization algorithms, the challenge isn’t computation—it’s coordination.
“Strategy and planning is the most important step in any healthcare implementation.” — Desai
Before AI can think, humans must agree on what it’s allowed to know.
Lemons: When Agriculture’s Data Runs Ahead
Travel to California’s groves, where agritech leaders like Limoneira are digitizing farm-to-table operations. Their AI story begins on conveyor belts, not in compliance meetings.
Every lemon is photographed six times; computer vision models detect ripeness and readiness for cold storage. Half a million units run through the system in a single session.
#Real Solution: From 15 Minutes to 15 Seconds
Lakshmi Duvulapalli built a Unity Data Catalog in Databricks as the backbone for an LLM-driven agent. Seasonal workers—often without computer access—use the AI Genie conversational interface on mobile devices to query data in real time. A process once requiring fifteen minutes now takes fifteen seconds.
Yet the mirror image of healthcare appears. Agriculture has real-time data, but it’s fragmented and fragile. Connectivity gaps between field sensors, cold storage, and distributors cause blind spots. Predictive planning relies on shifting weather models.
Agentic AI can predict and prevent spoilage only when systems share a common language. That is data amenability: harmonizing context across messy, moving realities.
What Both Sectors Reveal
Healthcare and agriculture reflect each other’s weaknesses.
Both suffer from real-time blind spots and legacy friction. AI fails not for lack of intelligence, but for lack of cooperation—between systems, teams, and timelines.
Lessons from MLOps and the Field
Technically, agentic systems thrive on continuous context. At Uber, 90% of machine-learning features are computed live, blending streaming and batch data for real-time personalization. In Tecton’s MLOps framework, shifting from distributed ML systems to unified ML canvases made large-scale context-aware decisions possible.
Whether routing a driver or a patient file, real-time context turns automation into augmentation.
The Human Factor in Data Readiness
AI fails when organizations chase models instead of meaning.
“AI projects don’t fail because of bad models—they fail due to misalignment with business priorities, weak execution, and poor governance.” — The Hard Truths of AI Adoption
Data amenability isn’t only technical—it’s cultural. Teams must share context, build trust, and co-own the systems that learn from them.
Or as AI Leadership Compass reminds us:
“If your people aren’t ready for AI, your data isn’t either.”
Agentic readiness begins with context engineering—the discipline of shaping what information an AI system can access, in the right sequence and scope.
“Context engineering is the art and science of filling the context window with just the right information for the next step.” — Andrej Karpathy
The Readiness Equation
Five signals of AI-ready supply chains:
Data Amenability Index: Interoperable, real-time, trusted data pipelines.
Human-in-the-Loop Maturity: Smooth decision flow between AI and operators.
Context Continuity: Insights available where work happens — field, warehouse, or clinic.
Risk Elasticity: Systems that adapt to regulation or weather without collapse.
Governance in Motion: Rules that evolve as data does.
From Lemons to Lifelines
Whether your product spoils in days or your data lives for decades, the same rule applies: agentic AI thrives on meaning, not models.
Healthcare shows that trust without agility limits action. Agriculture proves that agility without alignment breeds inconsistency.
True intelligence lies in preparing systems—and people—to cooperate with uncertainty.
In the end, turning lemons into lifelines isn’t about smarter AI. It’s about smarter readiness.
#DataAmenability #DataSupplyChain #FragmentedData, #FragmentedSystems, #AIReadiness #AgenticAI, #Healthcare, #Agrite #Tectonch
Dhaval Desai , Lakshmi Devulapalli
About the Author: Dr. Priya Sarathy, CDMP, is the founder of Wheel Data Strategies and author of AI Leadership Compass: Lead with Clarity – 7 Moves That Power AI Transformation. She advises enterprises on responsible AI strategy, data governance, and AI readiness for the age of autonomous systems.
