The thesis.

Claim one: physical systems will become observable

Over several decades, one category of activity after another has become measurable and then manageable through data. Financial transactions, logistics, software systems and communications all followed that course. In each case, continuous measurement replaced periodic inspection, and the organizations that adapted gained a lasting advantage in cost and reliability.

Physical equipment is among the last large categories where this has not fully happened. The control layer was digitized long ago. The condition layer, meaning knowledge of how healthy each asset is, was not. The components required to change that now exist at a cost that makes broad deployment plausible.

We think the direction is settled and the timing is the open question. AFFERENT is built on the view that the period of broad adoption is beginning.

Claim two: observation requires presence

Knowledge of a machine's condition cannot be inferred from a distance. It has to be captured at the machine, by a device in physical contact with it or close beside it. This is a meaningful structural fact about the market.

It means that software alone cannot solve the problem. It means that every installed device is a durable position: once a device is mounted, configured and trusted, there is little reason to remove it. And it means that scale is earned asset by asset, which is slow for everyone, including later entrants.

A business that requires hardware is harder to build than one that does not. We regard that difficulty as a feature of the opportunity. It is the reason the position, once established, is defensible.

Claim three: understanding compounds

The first device on the first pump can learn that pump's normal behavior. That is useful and limited. The thousandth device on the thousandth similar pump operates in a different situation. By then the system has encountered many instances of the same equipment across conditions, loads and ages, and has had the chance to observe how some of them degraded.

This is the mechanism behind the long-term thesis. Models of machine behavior are intended to improve with exposure. A wider footprint produces better models, better models make the product more valuable, and a more valuable product earns a wider footprint.

We are deliberate about how we describe this. It is a mechanism we believe in and are building toward. It is not a description of AFFERENT's present dataset. Every deployment has the potential to deepen our understanding of machine behavior, and that potential is realized only through deployments that actually take place.

Claim four: trust is the scarce asset

An operator who acts on a prediction takes a risk. Taking a machine out of service on the strength of a warning costs money if the warning was wrong. Ignoring a warning costs more if it was right. Operators therefore extend trust slowly and withdraw it quickly.

That has consequences for how a company in this field should behave. It should make conservative claims. It should explain its findings. It should report its own errors. A company that does these things consistently accumulates something a competitor cannot purchase: a record of having been right when it mattered.

The restraint visible across this website follows from that view. We consider credibility a strategic asset and manage it as one.

Where value accrues

In our analysis the hardware is necessary and the intelligence is where most of the long-term value sits. Devices can be imitated. An accumulated, validated understanding of how equipment fails cannot be assembled quickly by anyone, because it depends on time in the field.

The company is therefore organized around a hardware platform that establishes presence and a software platform that delivers continuing value: hardware and recurring intelligence software, as described under Investor Relations.

Risks

An honest thesis names what could prove it wrong.

• Execution risk. Building reliable industrial hardware is difficult, capital-intensive and slow to correct.
• Technical risk. Predictive performance must be demonstrated on real equipment at meaningful scale. It has not yet been demonstrated at that scale.
• Adoption risk. Industrial buyers are careful, sales cycles are long and trust is earned gradually.
• Competitive risk. Established industrial suppliers and other new companies are pursuing related opportunities.
• Supply and cost risk. Hardware businesses depend on component availability and manufacturing partners.
• Security risk. A security failure in a product attached to critical equipment would be seriously damaging.

This list is illustrative and incomplete. It is not a substitute for the risk disclosures that would accompany any formal offering.

The long view

Predictive maintenance is the first application because its value is direct and measurable. The larger outcome is a physical world whose condition is known: infrastructure that can be operated, maintained and planned on evidence. If that outcome arrives, the layer that connects computation to physical systems will be among the most consequential pieces of technology in use. AFFERENT's purpose is to build that layer.