Make or Buy: How to quickly test an AgTech Innovation?
The autonomous robotics market is taking shape. Every player is staking their position, announcements are multiplying, and one thing is becoming clear: those who don't get on board now risk missing the curve entirely. But the burning question remains: what should you build in-house, and what should you outsource or buy off the shelf?
The "I Can Do It Myself" temptation
In the automotive industry, no one is surprised that a manufacturer buys its transmission, steering system, or GNSS, these are commodities. The real question for autonomy is: what will become a commodity tomorrow? While the market is still finding its footing, the temptation lies elsewhere: "I have a handle on each individual technology, positioning, autosteering, certification, web development. Surely I can put something together myself." This is precisely where most disappointments begin.
The first question to ask yourself: where does your real value lie?
Before anything else, ask yourself a fundamental question: what is my core business, and where do I want to focus my efforts?
Running my factories and fleets at peak productivity?
Providing the best tools to support agronomic or operational practices?
Developing and commercializing a key technology that gives me a lasting competitive edge?
The answer shapes everything that follows. Building a complete autonomy solution — hardware, embedded software, web-based fleet management application, mobile control application — is already a daunting undertaking. But that's just the tip of the iceberg. What is consistently underestimated is everything that comes after: field support, operator training, ongoing maintenance, fleet-wide update management, regulatory documentation. These recurring costs are often heavier than the initial development costs, and infinitely harder to spread when you're starting from scratch.
The 4 risks you cannot ignore
Every autonomous machine development effort runs into four major risk categories, all of which are particularly difficult to budget for:
1. Value risk
Will customers perceive enough added value to buy and adopt your solution? A technically brilliant product that is poorly positioned or poorly documented will stay on the shelf.
2. Usability & ergonomics risk
Will your users be able to operate it effectively day to day? Field throughput — the surface covered or task completed per hour of actual work — is the first metric an operator looks at. A complex interface, frequent takeovers, or lengthy setup time directly reduce that throughput, and with it the perceived ROI. UX that isn't adapted to field conditions can sink an otherwise technically sound project.
3. Feasibility risk
Do you have the specialized expertise in-house? Robotics, computer vision, real-time processing, embedded cybersecurity, AI, safety certification… these skills are rare, slow to recruit, and expensive to retain.
4. Business risk
Will you be profitable in the long run? The cumulative hidden costs of staying current with the state of the art — hiring, R&D delays, retesting, unexpected integrations, ongoing maintenance — blow through initial budgets, even in the most disciplined teams.
The real challenge isn't technology
We often assume the main challenge of an autonomy solution is technological. In reality, the true challenge is twofold: first, service — providing support, training, and maintenance at fleet scale — and second, paradigm shift — changing how operators, agronomists, and site managers work. Both dimensions are systematically underestimated, and they are precisely what separates a great demo from a product that creates lasting value. The more complex and proprietary a solution, the harder these costs are to anticipate and control.
Make vs. buy in AgTech: the comparison
Criterion | MAKE (Build In-House) | BUY (Off-the-Shelf) |
Time to market | 3 to 6 years, depending on team maturity and use case complexity | A few months, rapid integration, operational testing in weeks, market-ready in under a year |
Technology ownership | Total, but must be built from scratch | Partial, vendor dependency |
Initial costs | Perceived as low… but misleading. Internal salaries seem less visible than licenses, but real costs accumulate fast | Higher and visible upfront, but manageable and predictable over time |
Hidden costs | Recruitment, turnover, delays, retesting | Intégration, formation, personnalisation |
Feasibility risk | High, rare skill sets required | Limited, expertise shared across multiple projects |
Safety & compliance risk | Variable by maturity, high for most; more manageable for those with certified product culture (Machinery Directive, ISO…) | Reduced, existing certifications, integrated field experience |
Scalability | Costly at scale, what works in demo may require deep rework to handle a real fleet, in both human and technical terms | Built in, designed to manage fleets from day one |
Differentiating value | Strong, if it's your core business | To be assessed based on positioning |
Field feedback | Zero at the start, long learning curve | Pooled across multiple customers |
Why buying structurally reduces your risk
Purchasing a solution already integrated across multiple machines means benefiting from the experience of those past integrations. Edge cases have been encountered, bugs have been fixed, certifications have been obtained. You're not starting from zero, and you learn fast.
The Make or Buy in AgTech decision comes with its own challenges. Buying involves risks such as vendor dependency, customization costs, and integration with existing systems. But building adds its own risks on top of all of those. The question isn't "Buy = risky" or "Make = controlled." It's: which path lets me best manage and cover my four fundamental risks while the market takes shape and standards emerge?
If your value is in the autonomy technology itself: consider building, but with eyes wide open about the skills and timelines required.
If your value is in the use case, the service, or the operation: buying frees you to focus on what truly differentiates you.
In either case, ask yourself: what will be a commodity tomorrow? What you're building in-house at great expense today may be available off the shelf in two years.
Our take
There is no universal answer. But one thing is certain: in every scenario, starting by purchasing an existing solution to validate your use case in real-world conditions is always the right move. It will cost you less and teach you far more than years of internal development, and yes, we're talking years, not six months. What you learn along the way will let you make an informed decision: Make, Buy, or a mix of both.
