Frameworks · POVs · Approaches

How we think

The theory behind the work. These are the frameworks and positions that shape how we approach novel problems — where the thinking comes from before the building starts.

01Framework

The Combination Gap

Most hard problems in industry are not unsolved because the individual technologies don't exist — they're unsolved because no one has combined the right ones yet. An AR overlay is trivial. An IoT sensor network is trivial. Connecting live sensor state to a technician's field of view in real time — that requires both, plus the integration thinking that most firms don't have.

What this means in practice: We start every project by mapping which disciplines are in the room and which are missing. The gap between what exists and what's needed is usually a combination problem, not an invention problem.

Cross-disciplinaryProblem framing

02Framework

Theory to Implementation Distance

There is always a distance between a working research prototype and something that survives contact with real operating conditions — temperature variation, network unreliability, operator error, maintenance cycles. Labs produce the prototype. Most firms deploy the prototype. We care about closing the distance.

What this means in practice: Every technical decision we make asks: does this hold at -40°C, does this work on 2G, does this survive someone who has never read a manual? The answer shapes the architecture from day one, not in the last sprint.

Engineering rigourProduction readiness

03POV

Network as Capability

A small lab cannot maintain deep expertise across AI, spatial computing, hardware design, and manufacturing simultaneously. What it can do is build and curate a network of specialists who trust each other and have worked together before. That trust is the actual product — it compresses the coordination cost of novel work dramatically.

What this means in practice: When you engage idezo, you are engaging the network. We know who is best for a specific chip architecture, who has shipped mixed reality in industrial settings, who can take a PCB from prototype to 500-unit run. That knowledge is not on any website.

NetworkCollaboration

04Framework

The Novelty Threshold

Not every problem needs a novel solution. Some need a well-executed standard one. We are specifically interested in work that sits above the novelty threshold — where existing tools, applied in existing ways, cannot produce the result. Below that threshold, there are better firms than us. Above it, there are very few.

What this means in practice: This is why we ask hard questions before starting. We are trying to establish whether the problem is genuinely novel or whether it only appears that way. Both answers are useful.

ScopeProblem selection

05POV

Spatial Computing is Not Screens

Most AR/VR deployments fail because they are built by teams who think in screens — flat interfaces repositioned in 3D space. Spatial computing requires thinking in environments: where is the user's body, what is their task load, what is the ambient lighting, what happens when they turn their head. The interaction model is completely different.

What this means in practice: We design for embodied cognition first, UI second. The question is never "where do we put the button" — it is "what should the user be able to do without looking away from their work".

AR/VRInteraction design

06POV

AI at the Edge

Sending industrial sensor data to the cloud for inference is fine until it isn't — latency matters when a machine is about to fail, bandwidth matters when you have 200 remote sensors, connectivity matters when you're 2000m above sea level. The most interesting AI work in physical systems happens at the edge, on constrained hardware, with real-time constraints.

What this means in practice: We design ML pipelines from the deployment environment backwards. What runs on a Jetson Nano, what runs on a microcontroller, what goes to the cloud, and when. That architecture is the product.

AIEdge computingIoT

If any of this resonates with a problem you're sitting on

The best conversations start before the brief is written.

Start a conversation See the work

Loading content...