Debono was built by pharmacy students. The core idea is simple: AI should sharpen the thinking that healthcare demands, not shortcut it. But in their current state, these tools are being used to answer questions that used to require years of clinical training and judgement. The primary ways that students of these professions interact with them all involve using them as a crutch, robbing them of their own development. AI integration doesn't need to be this way. The core urge that results in a student copy pasting something into that familiar light gray chatbox, isn't laziness, it is the path of least resistance. That is what the existence of Debono as a product seeks to prove.
AI is effective and reliable when its job can be confined to simple tasks. The approach of large tech companies is to improve a model so that it can handle… any task… reliably. I'm not qualified to comment on whether that direction is the right one, but I do know that in their current state, the models, when integrated intentionally, can be used to create so much more.
Healthcare is an interesting profession, and I believe that it is a great one for this experiment since, more or less, the same content must be learned by a field of different professionals with different backgrounds. Even
In homage
Edward de Bono
Thinking is a skill, not a trait.
Where did the name Debono come from?
The product is named after Edward de Bono, a Maltese thinker who spent his life arguing that thinking is a skill, not a trait. He was, by any reasonable measure, one of the most prolific minds of the late twentieth century. A physician by training, a Rhodes scholar at Oxford, and the author of over sixty books translated into nearly forty languages. He coined the term "lateral thinking" in 1967, and the phrase has since made its way into the Oxford English Dictionary, a fate that almost no living thinker gets to see happen with their own ideas.
What made de Bono's work stick was that he refused to treat thinking as something reserved for people who were already smart. His position was that the brain is a machine that makes patterns, and those patterns, once formed, become ruts. Even very intelligent people fall into their own ruts. The only way out, in his framing, was to apply tools, deliberately, that interrupt the pattern and force a new one. He gave those tools names. Six Thinking Hats. The PMI. Po. Provocation. Random entry. He treated them the way a surgeon treats a set of instruments, as things you pick up, use with intent, and put back down.
And he believed those tools could be taught to anyone. Schoolchildren. Boardrooms. Governments. In 1979 his thinking programme, CoRT, was adopted into the Venezuelan national curriculum, one of the few times in modern history that a country has ever taught thinking itself as a required school subject. He consulted for IBM, DuPont, and the United Nations. He spent the last decades of his life arguing, calmly and somewhat stubbornly, that the single most important human skill was also the one that was taught the least.
That is the person the product is named after. AI can hand you an answer, but it cannot hand you the thinking that got there. Pharmacy, like every clinical profession, is a discipline where the "arriving at" matters, the differential you ran, the drugs you ruled out, the interaction you almost caught. If a student skips the arriving at, the answer is useless even when it is correct, because the next case is going to demand they do the work without the AI's help.
So the name is a reminder, to us and to whoever uses the product, that the point is the practice, not the output.
Roadmap
The overall roadmap
Three versions. Three gates. Each one means something specific about what the product can do.
- 1.0 Shipped
April 2026
A good study tool
Debono 1.0 fulfills the role of a good study tool well. As a tool dedicated to the pharmacy curriculum, it fulfills its purpose well.
- The data pipelines are in place and stable.
- Users now fuel development rather than test the infrastructure.
Supporter detail- Pharmacy was chosen first because we had the domain experts and the grader in house
- Dual RTX 3090 inference running identical 8B models in a round robin
- Context aware explanations that cite the student's own slides, not a generic textbook
- 1,100+ drug knowledge graph mapped against MOAs, interactions, and indications
What shipping 1.0 actually meant
Not "it works for the board exam." It meant a student could live inside the product from enrolment through licensure, and the product kept up.
Admin notes- Public launch at debono.ai. Onboarding funnel instrumented through Umami and internal events.
- Next 6 weeks: conversion tuning, Itachi E2E coverage expansion, referral mechanic.
- 2.0 Next
Q1 2027
Users come back for the game
The point at which users come back for the game rather than the pressure of an impending exam.
Supporter detail- Medicine is shipped as a class at the same depth as pharmacy. Not a second product, a second class in the same leveling zone.
- Nursing is shipped the same way. Three professions living on one platform is what lets the word 'classes' actually mean something.
- The mobile app is shipped. For most students the phone is the real study surface, so the experience has to be built for it, not adapted to it.
Why the concurrent attack
2.0 is not one launch, it is three. The game has to be good on its own, medicine and nursing have to exist as classes at the same quality bar as pharmacy, and the mobile app has to ship. Any one of them alone would be a point release. Putting them together is what earns a major version.
Admin notes- Blocker: medical content partnership. Scoping active.
- Architecture dependency: class slot refactor in 1.9 blocks 2.0. See intra 1.x roadmap.
- 3.0 Vision
2028
All of healthcare
When Debono becomes a place where people from all backgrounds of healthcare feel like they have a tool made specifically for them and have the ability to collaborate.
Supporter detail- Cases are co authored by real hospital teams who have run them on actual patients
- Entry requires prior proficiency in at least one class before a student is allowed into the raid
- Post raid review mode lets the team replay every decision and see what changed the outcome
- Live instructor observation mode for preceptors and attendings
- Team ranking meta where cohorts climb a shared board by case quality, not volume
The bet
Interprofessional education has been a twenty year conversation that has not actually shipped anywhere at scale. We think the raid mechanic is what finally makes it land, because the scoring is the carrot and the team is the content.
Roadmap
Between 1.0 and 2.0
The minor version bumps that get us from one class to many. Hover any card for the reveal.
- 1.1 In progress
Q3 2026
PvP and guilds
Supporter detail- Timed question duels
- Global and school leaderboards
- Guild chat for study cohorts
- Daily challenge rotations
- ELO style matchmaking scoped to NAPLEX domains
- Supporter exclusive mentor role in guild chat
Perk
Supporters get mentor flair in guilds and early access to each new PvP season.
Admin notes- Blocker: moderation tooling for guild chat before launch.
- 1.2 Next
Q4 2026
Instructor mode
- 1.3 Planned
Q1 2027
Mobile first
- 1.4 Planned
Q2 2027
Expansion content
- 1.9 Planned
Q4 2027
Class slot refactor
Supporter detail- Generalise the class schema so a second profession can be added without rewrites
- Shared profile and progression system across classes
- Cross class navigation and visibility
- Class agnostic question engine
- Private preview of the second class scaffolding before public launch
Why this version exists
Shipping a second class without this refactor would double the code paths and halve the velocity. 1.9 is the bridge that makes 2.0 cheap.
Comments
Get an email when someone replies on this post.