Platforms taken from idea to production

The business situation

Nexanomics engaged me to take a business idea from concept to a market-ready business unit: defining the market, proving there was a real problem worth solving, designing the commercial model, and building the platform to deliver it.

The problem is one almost every organization with a website has. People find them, then drop away somewhere between finding them and choosing them, and the organization can't see where, why, or what it costs. They already have analytics, but analytics are data, not answers. A café owner, a physiotherapist or a charity director has no marketing team to interpret dashboards and no budget for a consultant charging thousands a month, so nothing changes and the cost compounds month after month.

What was at stake

The research defined the market by situation rather than industry: any organization that has a website, wants to grow, and has no marketing team. That is a market measured in tens of millions of websites across the major English-speaking markets. The gap was clear: plenty of tools report what happened, and none tell a small business what to fix first at a price it can accept without a budget debate.

The risk was the usual one for a new venture: a multi-tenant platform normally needs a cross-functional team and around two years before the first customer, with a great deal of capital committed before the market has had its say.

What I did

• Market and proposition. Defined the customer, sized the market, and shaped the proposition around "intelligence, not analytics": every subscriber receives a monthly A–F Scorecard across Discovery, Website and Social, and a Fix-Guide ranking what to fix first.
• Commercial model. Tiered subscriptions paid upfront quarterly or annually, so cash arrives before costs; early-adopter pricing; a built-in referral mechanism; and a partner channel that gives digital agencies a recurring, pre-researched list of billable improvements for each client.
• An AI-native platform. AI agents audit each subscriber's customer journey every month, and a quote agent collects supplier quotes for fixes that need outside help. The owner approves anything that commits money or sends a message. The platform is natively multi-AI, running live across three AI providers.
• The build. Architecture, security, more than fifteen API integrations, cloud infrastructure in secure regional zones in Australia and the United States with data kept in-region, billing, group and franchise models, board papers and full production documentation, built with AI accelerating every stage and each release signed off through a documented acceptance walkthrough.

The results

A market-ready platform with its commercial model, channels and operating documentation in place, built at a fraction of the usual cost and time. The business owns everything it created: the platform and its intellectual property, the commercial model, and the documentation to run and extend it.

Proving the market. Proof of market is designed in rather than assumed: beta operators, early-adopter pricing that rewards committing early, direct sales to build the first case studies, and live operation in two regions. This website is itself a live customer in the United States zone.

What it means for your business

This is how a new business unit should be launched inside any organization: define the market and the problem first, design the commercial model with the product rather than after it, lean on AI to compress the build, and prove the market before scaling the spend. On a client engagement, your developers become the builders. Visit Nexanomics.

The business situation

The business set out to make personalized, one-to-one teaching available to students who can't otherwise get it: students who are remote, isolated or out of normal schooling for a time, such as those in hospital. The aim was to reach them, not to replace classroom teachers. The question was how to give thousands of students what feels like a private lesson, all at the same time.

What was at stake

Trust in AI in education is low, and falling. The common perception is that AI in the classroom means "leave it to the AI": students left alone with a tool, and no teacher directing the learning. Almost all education technology reinforces that picture, because it is built around self-directed learning products, where the student finds their own way through content.

This business took the opposite position. Its system is founded on teaching: a strongly directed, human-derived, interactive teaching process in which educators structure and author every lesson, and the AI delivers it one-to-one within the scope they set. The teaching is delivered by a lifelike digital human, so every student has a teacher in front of them rather than a screen of content. Designing that for massive scale, with every session individual, was truly novel. It meant real technical uncertainty, a multi-year investment that had to be structured so the business could fund it, and, as the work showed, a second hard problem: the cost of creating good course content in the first place.

What I did

I designed and led a cross-functional team of eleven to build a human-structured, teacher-led teaching ecosystem rather than a single product:

• A teaching platform. Using digital-human technology, a digital teacher delivers structured, educator-authored lessons in individual one-to-one sessions, to any number of students at once, adapting its delivery to each student's learning style. Educators set the guardrails, and in-lesson tests are marked as the lesson runs.
• A course-authoring platform. AI drafts a course outline from existing material such as curriculum documents, lesson plans and video; a subject expert reviews and shapes it; then lessons and slides are built and validated before publication, along with subject guardrails and answers to common student questions.
• A course marketplace. A design for publishing courses by subject, age level and curriculum, so expert-authored content can reach schools and learners at scale.
• An AI-driven development system. Reference documentation paired with machine-readable context, so AI coding sessions can build correctly without re-briefing. It cut build cost and delivery time by roughly 75% each.

From the start, the work ran as a formal R&D program: each technical unknown was framed as a documented experiment, with core activities kept distinct from supporting ones. I worked alongside the business's specialist R&D advisors, accountants and legal counsel.

The results

Working teaching and course-authoring platforms on a shared, secure multi-tenant foundation with more than a dozen third-party integrations, and a development method that now builds at around a quarter of the previous cost and time. Around $30M of eligible R&D was claimed over three financial years, returning $13.5M through the R&D Tax Incentive.

What it means for your business

Three lessons carry over to any new initiative. Where trust in AI is low, keep people visibly in charge and let AI extend what they do. Solve the bottleneck the market actually has, which is not always the one in the original plan. And if you're building something new, design the build so it can help pay for itself from day one. How R&D-ready builds work.