Criteria first
Pass conditions are defined per test scenario before testing begins.
method
How a POC actually runs
A proof of concept is worth running only if both sides will accept the answer before they know what it is. The method frames the unknown, fixes pass criteria in advance, isolates risk and reports whichever way the results fall.
how it runs
Five stages, from need to decision
Each stage carries context forward so the startup, partner and test team arrive with one shared definition of success.
- 01
Operational need
The partner states the operational problem. We frame it into something testable.
- 02
Global scouting
Once the need is framed, we search globally, assess candidates technically and put a short list in front of the people who will host the test.
- 03
Partner match
The startup and partner define the question, test and criteria.
- 04
Field POC
Integration, instrumentation and testing happen in the environment where the technology has to perform.
- 05
Scale what works
The partner decides whether to take it further, test again with a changed scope, or stop.
what a POC produces
A written answer, against criteria agreed in advance
Before anything is built, both sides write down what success looks like. The final report states objectives, setup, test plan, results per scenario, conclusions and recommendations.
Real environments
Testing can take place in industrial facilities and operating environments appropriate to the question.
An answer either way
Results that do not support a rollout are reported as clearly as results that do.
Resolution
- Scale
- Reconfigure + retest
- Useful no
the kind of problem we work on
Representative challenges
These are examples of the kind of operational need we frame and test against. They are not open calls.
These categories describe the kind of work we do. They are not current opportunities, they carry no deadline, and applying against one does not create a live process. When we run an open call we will say so explicitly and date it.
Representative — not an open callLogistics
Warehouse and last-metre logistics automation
Storage across five or six levels with two- and three-deep racking, picked by hand with a ladder. Or a package that has to travel the last metre from a vehicle, through a door, into a lift, to a floor. We have tested both shapes, including autonomous ground robots doing indoor and outdoor delivery runs.
Representative — not an open callAutomotive and mobility
In-vehicle experience and software-defined vehicle applications
Communication between a vehicle and the road around it. Personal audio zones for driver and passenger. An assistant that answers from the owner's manual. These are cabin problems, and they are only solved when the technology disappears into the vehicle.
Representative — not an open callLogistics and heavy transport
Fleet and driver safety, and predictive maintenance
Large vehicles have zones the driver cannot see. Vehicles get damaged during loading and transport. Maintenance happens after the failure. Each is measurable, each has a cost the operator already knows, and each is testable on a real fleet rather than in a lab.
Representative — not an open callEnergy
Energy efficiency, alternative fuels and hydrogen
Load prediction and distribution that accounts for renewables. Fuel substitutes and low-carbon alternatives. Hydrogen storage and handling. Energy as a service where there is no capital to spend up front. These are tested in plants that cannot stop running while you test them.
Representative — not an open callManufacturing and materials
Sustainable industrial materials
Replacing a preparation step in an existing coating line. Running a wood-waste material through a plastics process. Turning a refinery waste stream into construction aggregate. Materials work is slow, it is graded by outside laboratories, and partial passes are common.
Representative — not an open callIndustry 4.0 and energy
Inspection and robotics in hazardous environments
Leak detection without sending a person. Inspecting pipe condition under insulation. Mapping infrastructure that is buried. Flying or driving a robot somewhere an explosion risk is real. The technical question and the certification question have to be answered together.
Representative — not an open callEnergy and industrial services
Acoustics and noise control in industrial facilities
Silencing baffles have barely changed in fifty years. They are costly, they take space, and in hundreds of installations a year they are the default answer to a regulatory limit. Both an active-cancellation route and a materials route are worth testing.
Representative — not an open callCross-sector
AI for operational knowledge and decision-making
A refinery's technical and HR manuals. A service organisation's diagnostic knowledge, sitting in the heads of people who are retiring. Operational chatter across radio, messaging and email that nobody has structured. The test is always the same: ask it things you already know the answer to, and count.
Representative — not an open callData infrastructure
Data-centre and mission-critical facility operations
Floor space lost to cages and drywall. A technician who has to physically travel to a site before anything can be fixed. Human error during routine maintenance. Corrosion on external cooling pipework in coastal air. These are narrow, expensive and very testable problems.
test capability
A workshop, an instrumented vehicle, and working sites
The workshop supports integration and bench mockups. An instrumented Kia EV6 provides a vehicle platform. Partner environments support tests that cannot be simulated.
KIA EV6 / TEST PLATFORMINSTRUMENTED
Integration, Isolated, Risk control
reporting
One report format, whatever the result
Executive summary, objectives, setup, test plan, results per scenario, conclusions and recommendations. Results are stated against the criteria fixed before testing.