Renvar Labs
Nine programmes. Each carries a complete specification — architecture, scope, timeline, budget, risk analysis, go-to-market and IP strategy — and each is sequenced behind the platform capability or market signal it depends on.
Why publish them
Because the products we ship are steps toward these. Two of the original eleven programmes have already moved into the portfolio.
Cluster 01
Capital-intensive, closer to hardware, and contested by well-funded incumbents. Renvar works these because they are the substrate the remainder of the portfolio will eventually require.
3 programmes
Intention as an input device.
Spatial computing still waits for hands. This track investigates neural and biosignal input as a first-class channel into AR/VR environments, behind the same signal abstraction the personal agent already uses — so a new source of intent is an implementation, not a rewrite.
Moat · Signal abstraction + interaction patents
Simulation accurate enough to plan against.
Synchronised digital twins of physical systems — buildings, fleets, feeders — that stay faithful enough for an agent to plan against them and cheap enough to run continuously rather than as a consultancy exercise.
Moat · Data fidelity + integration depth
Inference where the sensor is.
Personal memory that cannot leave the device and robots that cannot wait on a round trip both point at the same requirement. This track works on the inference platform for that constraint, including neuromorphic silicon as a long-horizon path.
Moat · Hardware/software co-design
Cluster 02
Limited competition, substantial latent demand, and the structure of an infrastructure layer rather than an application. Strategically the more significant half: layers are what endure over a five-to-ten-year horizon.
6 programmes
Settlement infrastructure for agent-to-agent transactions.
AICOPay solves bounded spending for one agent and one person. The general problem is a network: discovery, delegation, settlement and dispute resolution between autonomous parties. The prize here is a standard, not an application.
Moat · Standard + network effects
Authenticity in a world of cheap synthesis.
Human-or-machine attestation, anti-voice-cloning, provenance of media and of claims — delivered as a layer other products embed rather than a destination anyone visits. Already surfacing inside AICO's conversations and VITRIUM's wall labels.
Moat · Trust + regulatory alignment
Demographic change is the market.
Eldercare technology that treats the family as the unit rather than the patient record: a single agent extended across people caring for one another, with certification requirements addressed from the outset rather than retrofitted.
Moat · Demographics + certification
Neutral custody for personal context.
Sovereign, encrypted, portable personal memory, and the commerce model that follows from it: on-device matching in which the agent is aligned with the individual and the revenue share accrues to them rather than to an advertiser. Feeds directly into Core Memory.
Moat · Neutrality + portability
Accountability for autonomous decisions.
Audit and insurance for autonomous systems: what an agent did, under whose authority, and who bears the loss when the decision was wrong. Regulatory standing is the defensible position, and MEXUM's expert sign-off model is the first operating instance of the pattern.
Moat · Regulatory status
Robot foundation models are constrained by the scarcity of real-world data.
Capturing manipulation and navigation data from robots operating in the world, structured for training. Every Embodied Agent OS deployment is a potential collection point, which makes the data moat a consequence of the product rather than a separate business.
Moat · Data accumulation
Embodied Agent OS and Grid Edge Orchestration originated here. If one of these is the layer your organisation is missing, that is the conversation we would like to have.