Daniel Gibbon once had a beautifully legible plan. Finish school near the top of Queensland. Study engineering at the University of Queensland. Continue through a provisional pathway into medicine. Build artificial hearts or biomedical implants. It was the kind of plan that makes relatives relax at dinner. It had grades, prestige, and a destination that could be explained before dessert.
Then he applied an engineer's least romantic habit to his own life: he ran the comparison. Gibbon kept returning to a question that is awkward precisely because it is so short: excellent relative to what? Medicine offered meaningful work. Yet he became interested in the system upstream of every surgeon, researcher and inventor. Each began as a student. Each had once depended on a teacher with limited time, uneven resources and another stack of assignments waiting at home.
That thought did not produce Edexia in a flash. First it produced tutoring. At 18, while earning a reported 7.0 out of 7.0 university GPA, Gibbon co-founded 99 Plus Education with Nathan Wang. They built STEM courses, resources and tutoring around mastery-based learning, grew a team of more than 30 tutors and served hundreds of students across Queensland. Y Combinator would later describe the business as reaching A$200,000 in annual recurring revenue in a little over a year.
The useful first company
99 Plus looked like success, especially for two teenagers building between lectures. It also gave Gibbon a better education than a spreadsheet could show. He saw the difference between a student who wanted to learn and one who merely wanted the next answer. He saw that teachers already understood feedback, motivation and progression. The shortage was not imagination. It was hours.
Gibbon and Wang handed the tutoring company on and started asking a broader question about education. They met teachers and researchers, reportedly holding hundreds of conversations. Their first year at Edexia, founded in 2023, was less a product launch than an apprenticeship in somebody else's workload. They explored before they narrowed. By the time they launched an assessment product in August 2024, the starting point was almost stubbornly ordinary: marking.
“The biggest pain point was teacher workload.”Daniel Gibbon
Marking is an odd candidate for reinvention. It happens quietly, one page at a time. It is repetitive without being mechanical, because a score can alter how a student understands their own ability. Useful feedback requires attention. Consistent grading requires judgment. The more carefully a teacher does it, the more evening it consumes.
A rubric is compressed experience
The easy sales pitch for AI grading is speed. Feed in an essay, receive a score, watch a weekend reappear. Gibbon's product argument is more careful. A general model does not know what a particular teacher means when a rubric asks for a “discerning evaluation.” Two experienced markers may read that phrase differently. Their standards contain examples, local context, subject knowledge and years of pattern recognition. Much of it is known on sight and difficult to dictate.
Edexia therefore treats the rubric as a conversation rather than a commandment carved in stone. The software shows how it has interpreted the criteria. A teacher can correct the output, refine the model's understanding and moderate the result before it reaches a student. The machine drafts; the teacher remains responsible. It is a product architecture built around the fact that professional judgment does not fit neatly into a prompt box.
The teacher-controlled assessment loop
This is where Gibbon's technical and educational instincts meet. He studied software engineering, but his first businesses put him near the people using the software. He speaks about ambiguous rubric language as an alignment problem. The phrase sounds like laboratory equipment. In a classroom it means something practical: can a system absorb a teacher's standards without flattening every good essay into the same obedient shape?
There is a second tension. A tool that helps students too eagerly can remove the thinking school is meant to develop. Gibbon has publicly argued for a more Socratic direction, where AI asks questions instead of supplying the finished response. He has also argued that teachers need visibility into how students produced digital work. The company is trying to occupy a narrow strip of ground: use the technology, preserve the struggle that makes learning useful, and keep an adult accountable.
The launchpad and the feedback loop
Early school partnerships moved quickly. By late 2024, Edexia was working with 12 Queensland schools, including Brisbane Grammar, St Margaret's and Gregory Terrace. In November, Gibbon and Wang announced that Edexia had been accepted into Y Combinator's Winter 2025 batch. They had larger offers from Australian venture firms, Gibbon said at the time, but chose YC as a better launchpad for the founders and the company.
San Francisco offered an unusually dense community of people living inside similarly unlikely plans. Gibbon described a routine of long workdays punctuated by YC events and conversations with founders from around the world. The change he valued was social as much as financial: other founders understood the peculiar combination of ambition, uncertainty and repetitive labor.
“The founder journey can be brutal.”Daniel Gibbon, on finding community at YC
The program also sharpened the company's preference for evidence over startup theatre. “YC forces you to focus on what truly moves the needle,” Gibbon later said. For Edexia, that meant school contracts, working deployments and short feedback loops with teachers. By 2026, an alumni account from Gibbon's former school reported more than US$600,000 in school contracts secured around the YC period and US$4 million in equity funding.
Money enlarged the experiment; it did not settle it. Edexia narrowed toward one of assessment's less forgiving corners, senior-secondary English. An English essay resists easy automation. It can be technically competent but lifeless, unconventional but perceptive, polished but empty. Gibbon said the company worked with experienced assessors to calibrate performance across Australian curricula. The same product that once promised subject-agnostic grading began concentrating on the place where nuance could not be politely ignored.
A founder with a notebook
Gibbon describes himself as a “life optimiser,” an alarming phrase until he explains it with the homely metaphor of sharpening a saw. He likes routines, logical decomposition and the small machinery of habits. Atomic Habits helped him build the foundations. He has worked punishing schedules, but he also talks about sustainability, walks and making time for close relationships. The optimisation is meant to extend the effort, not simply decorate exhaustion.
There is a pleasing analogue streak beneath the AI founder. He still plans with pen and notebook. After hearing a teacher argue that typing can help creativity because ideas can be rearranged and expanded non-linearly, he publicly reconsidered the usual classroom argument for handwriting. He did not convert on the spot. He simply admitted that the question had become less tidy. It is a small example of his recurring method: state a belief, find an inconvenient comparison, inspect the machinery.
The larger comparison remains the one that redirected his career. Gibbon's stated ambition is a world where education produces curiosity, critical thinking and purpose rather than mere obligation. He believes teachers should have more time to mentor, inspire and connect. Those are enormous claims to place on a grading assistant. Yet the point of the assistant is to make a little room for them.
Education technology has a long habit of arriving with a trumpet and leaving teachers with another login. Edexia's fate will depend on whether it can avoid becoming one more system to manage. Its defensible idea is not automation alone. It is deference: the teacher trains, inspects and decides. The software must earn trust one correction at a time.
For Gibbon, the red pen is an entry point rather than a destination. Assessment reveals what a school values, what a student understood and where a teacher's time goes. Improve that loop, and perhaps feedback arrives sooner. Perhaps practice becomes more frequent. Perhaps the teacher gets back an evening. The company has chosen a modest object for an immodest ambition.
The next phase reaches beyond long essays. In August 2026, Gibbon described short-answer exam tools that let departments assemble activities and share questions across classes. He also said Edexia was developing teacher-approved feedback workflows in more than 45 Australian schools. The numbers matter less as decoration than as exposure to difference: every new department brings another interpretation, another exception, another reason the rubric cannot be treated like a vending-machine instruction. Scale, in this business, means collecting more nuance without losing the person responsible for it.
The laminated life plan is gone. In its place is a stack of essays, a difficult rubric and a question with no final answer: excellent relative to what? Gibbon has made a company out of continuing to ask.