Building systems that increase human capability

The thesis underneath the work

There is a question that sits at the centre of every product decision Mustard Seed Group makes. It is not "can AI do this?" It is not "how do we automate this?" It is something older and more demanding: does this make the person more capable?

That question is the organising thesis of this company. It is what connects a fitness coaching platform to a B2B operating system to a consumer media ecosystem to a research lab. From the outside, the portfolio can look scattered. From the inside, the line is clear. Every product in this group exists to increase the capability of the person using it: their physical capacity, their commercial intelligence, their creative output, their ability to understand and navigate complexity.

We call this the capability thesis. It is not a marketing position. It is a constraint we apply to product decisions, and it regularly tells us what not to build.

What capability actually means

The word capability is easy to say and easy to hollow out. So it is worth being specific about what we mean.

Capability, as we use it, refers to a person's actual range of effective action. Not their confidence. Not their access to information. Their ability to do something real in the world that they could not do before, or could not do as well.

A system increases capability when it extends what a person can accomplish, at a higher level, with their own judgment still in the loop. A system that does something for you is not the same as a system that makes you better at doing it yourself. The distinction matters more than it sounds. The first type creates dependency. The second type compounds over time.

This does not mean automation is bad. Automating the repetitive so the person can focus on the consequential is a genuine capability increase. But the test is always: after using this system, is the person more capable, or have they just offloaded something they can no longer do without us?

We build toward the first. We are sceptical of the second.

The four registers of capability in the portfolio

The portfolio spans four distinct registers of human capability. Each one is a domain where Mustard Seed Group believes systems can make a meaningful and durable difference.

Physical performance

CheekyGains is a consumer fitness and performance platform. Inside it, Naira operates as an AI performance coach: one that adapts to the individual, responds to how they are actually training, and builds programming that evolves with the athlete.

Physical performance is the most concrete register of capability. You can measure it. You can track it across weeks and months. And the feedback loop between effort and result is immediate in a way that most knowledge work is not.

What makes this register interesting is that it requires the system to respect human variability in a deep way. A coaching programme that works on average is not useful to someone whose recovery, schedule, and history is their own. Naira is built around the premise that good coaching is specific, that specificity requires genuine responsiveness, and that the system should increase an athlete's understanding of their own body and performance, not just their compliance with a plan.

The long-term goal is not to create people who need Naira to train. It is to create people who train better because Naira helped them understand how.

Commercial execution

Orbit is a B2B SaaS operating system built for the lead-to-launched-product workflow. It is the product of repeated observation of the same failure mode: organisations that have talent and ideas but cannot translate them into consistent commercial output.

The gap is rarely intelligence or effort. It is usually systems. People spend time and cognitive energy on coordination, tracking, handoffs, and prioritisation that should be handled at the infrastructure level. They make decisions about what to work on next that should be made once and updated automatically. They lose context in transitions that should be seamless.

Orbit is designed to close those gaps without turning the people inside it into administrators of their own work. The system should carry the coordination overhead so the people can carry the thinking. Orion, the AI intelligence layer that powers Orbit, is built to augment the judgment of the operator, not to replace it. It surfaces what matters, flags what needs attention, and does the computational work of keeping everything coherent, so the person can do the higher-order work of deciding what coherence should look like.

TUXX, the custom AI systems and software studio behind many of the group's most experimental bets, proves these patterns commercially. Pattern Up is one of its sub-products. The work at TUXX is about taking the logic developed across the portfolio and building purpose-fit systems for specific commercial problems, quickly, without losing quality.

Creative and cultural work

All Purpose is a consumer app and media ecosystem. Its sub-products include All Purpose Music, Relay, Horizon, and Made It Out. Each exists in a different part of the creative and cultural space, but they share an orientation: the person at the centre of the creative process should be more capable because of the tools around them, not less.

This register of capability is the most contested right now. There is enormous pressure in the creative industries to use AI as a replacement for creative labour rather than as an amplifier of it. The economics are tempting. The logic is straightforward. But we think it is a short-term view.

Creative capability is not just the ability to produce output. It is the ability to make decisions about what is worth making, to develop a distinct sensibility, to build an audience that trusts your judgment. None of that is automatable in any meaningful sense. What can be supported is the execution layer: the distribution, the discovery, the administrative weight of getting creative work into the world. All Purpose is built around the idea that clearing that weight is what allows creative capability to compound.

Research and intelligence

Benediction Lab is the research arm of the group. Its work covers agents, memory, GUI control, and autonomous product development. It is early-stage by design. The questions it is working on are not product questions yet; they are infrastructure questions that will shape what is buildable in two or three years.

The intelligence layer connecting research to product is Orion. As Benediction Lab develops findings, those findings inform how Orion operates, what it can reason about, and how it handles the kinds of tasks that currently require constant human oversight.

This register of capability is perhaps the least visible to an outside observer, but it is the most foundational. The degree to which a person can understand, navigate, and act on complex information is not fixed. It is shaped by the quality of the intelligence infrastructure around them. Building that infrastructure well is a long-term project, and one that requires genuine research rigour rather than just product iteration.

What makes this a holding company, not a fund

There is an important structural distinction that shapes how Mustard Seed Group operates. This is not a fund. A fund deploys capital into companies and manages for portfolio-level returns, with each company operating independently toward its own outcome.

MSG is a holding company with an organising thesis that runs across the portfolio. That means the products inform each other. Research from Benediction Lab shapes Orion, which powers Orbit, which is built by teams that work across TUXX. The consumer behaviour insights from All Purpose and CheekyGains inform how we think about motivation, retention, and the design of capability-increasing systems across the whole portfolio.

The integration is not accidental. It is the strategy. The thesis only becomes durable if the knowledge generated in one part of the portfolio can be applied across the others. A fund optimises each company in isolation. A holding company with a coherent thesis builds compounding advantage across the whole.

This also shapes how the group approaches talent. People who work across MSG are not generalists in the vague sense. They are people who can apply a consistent set of principles to different domains and learn from the differences. That cross-domain learning is a significant part of what the group produces.

Why a broad thesis is more durable than a narrow one

The obvious objection to the capability thesis is that it is too broad. A company that builds fitness platforms and B2B operating systems and media ecosystems and research labs looks, on the surface, like it lacks focus.

The answer is that the breadth is intentional, and it is what makes the thesis durable.

Consider what happens to a company organised around a narrow thesis: "we make AI-powered CRM software" or "we build fitness apps using computer vision." When the technology shifts, when the category saturates, or when the problem changes shape, the company has to rebuild its identity from the ground up. The thesis is too dependent on a specific technology or a specific market configuration.

A thesis organised around a persistent human need is different. The need for capability increase across physical, commercial, creative, and intellectual domains is not going away. The specific tools that serve it will change. The registers in which it operates will evolve. But the underlying logic, that people and organisations want to do more, understand more, and achieve more with the resources they have, is stable across generations.

The portfolio is broad because human capability is broad. The thesis is focused because the principle is singular.

What MSG is building toward

The honest answer is that the endpoint is not defined in precise terms. This is a deliberate choice, not an evasion. Companies that define their endpoint too early tend to optimise for it too literally and miss the actual opportunity as it develops.

What we can say is this: Mustard Seed Group is building toward a world where the systems around a person, whether they are an athlete, a founder, a creative, or a researcher, genuinely increase what that person can do. Where the infrastructure handles the weight that should be handled by infrastructure. Where the intelligence layer makes the human judgment inside it sharper, not redundant.

That is not a simple product to build. It requires research into how agents work at a fundamental level. It requires commercial discipline in how those capabilities are packaged and delivered. It requires consumer insight into how people actually adopt tools that are supposed to help them. And it requires a willingness to say no to capabilities that are technically impressive but do not actually serve the thesis.

The portfolio will grow. Some of what we build will work better than expected. Some will need rethinking. But the question at the centre of every decision will remain the same: does this make the person more capable?

If the answer is yes, we build it. If it is not clearly yes, we ask harder questions before we proceed.

That is the thesis. That is the work.