The first meeting often starts with a product video and ends with a promise to send a quotation. That is not useless, but it is the least demanding part of the evaluation. Agricultural robotics has to handle variation, repetition, maintenance, safety, and the commercial realities around a crop or facility.
Use the first conversation to make the company's assumptions visible. You are not trying to punish an early-stage product for being early. You are trying to understand what kind of decision you are making.
1. Define the decision
Are you looking for a finished machine, a component, an integration partner, a manufacturing partner, or an engineering team? These are different searches with different evidence requirements. A strong component company may not be the right finished-product vendor. A capable integrator may not own the core technology.
2. Ask for the operating envelope
Request specifics: crop or material, variety, geometry, light, terrain, temperature, humidity, working speed, operator involvement, and the conditions under which the system stops. Ask what the company has not tested yet.
A useful answer can include limitations. A vague answer is often a sign that the product is still being defined or that the sales conversation is ahead of the engineering evidence.
3. Separate the standard product from the demo
Which parts are standard and repeatably available? Which sensors, software modules, grippers, chassis, or fixtures were built specifically for the demonstration? How many units exist? What is the current production lead time?
4. Test commercial maturity
Commercial maturity is more than a number of units shipped. Look for repeatable commissioning, maintenance procedures, spare parts, software updates, operator training, warranty terms, and a person responsible for the customer after delivery.
If the product is early, ask what the company is trying to learn in the next deployment. A pilot is more useful when both sides agree on what evidence will change the decision.
5. Check export readiness
- Documentation, interface descriptions, manuals, and training in a usable language.
- Software, cloud, data, cybersecurity, and remote-support dependencies.
- Electrical, radio, machine, chemical, aviation, or other market-specific compliance needs.
- Spare parts, consumables, batteries, packaging, shipping, commissioning, and local service.
- Ownership of designs, source code, improvements, field data, and customer-specific integration work.
6. Make the visit prove something
A factory visit should not be a tour of rooms. Bring a short list of questions, request a relevant demonstration, and watch the handoffs: who explains the system, who operates it, who diagnoses a fault, and who can discuss the next engineering step?
When possible, visit a working customer or test facility, not only the showroom. Observe the system after the first impression has worn off.
Use a simple decision memo
After each meeting, record what was demonstrated, what was claimed, what was observed, what remains unverified, and what would be required for a pilot. This makes comparison fairer and prevents a fluent presentation from carrying more weight than field evidence.