Every learner gets a class of one.
Eleven ways Zavmo personalises a lesson, what each one changes for the person being taught, and what four expert testers said about being taught that way.
Most platforms teach a topic. Zavmo teaches a person.
That reads like marketing until you look at what it costs to mean it. The system has to find out where you actually are before it teaches you anything. It has to know enough about your working life to draw its examples from it. It has to notice when you have stopped following, and change what it is doing, without you having to admit you are lost.
Forty years ago Benjamin Bloom showed that a learner given one-to-one tutoring with mastery progression outperforms a learner in a conventional classroom by two standard deviations. That is the gap between an average student and one in the top two per cent. He asked, in print, whether anyone could find a way to give every learner what only the wealthy had ever been able to afford. The paper has been cited more than six thousand times and the system changed not at all, because nobody could afford thirty teachers for thirty children.
That is the constraint Zavmo exists to break. This page is about the only place where breaking it counts: what it is like to be the learner.
The strongest positive themes relate to the platform's personalization, relevance to the individual learner, conversational AI experience, and its potential to create a more transformational and action-oriented learning journey than traditional online learning platforms.
Programme director, university partnerPilot feedback report, consolidating four expert testers, August 2026The same learning outcome, taught two ways.
Personalisation on most platforms means being routed down one of three pre-made tracks. On Zavmo the tutor's entire brief is rebuilt from scratch, every turn, from everything the system currently knows about you. Two learners covering the identical assessment criterion do not get the same lesson at different speeds. They get different lessons.
Learner A and Learner B
"Risk assessment is the process of identifying hazards, evaluating their likelihood and severity, and determining appropriate controls. Consider the following case study."
Identical text, both learners.
Learner A, junior IT auditor. Learner B, operations coordinator.
"In your role as an auditor, a control gap you find in week one is a very different conversation from one you find at sign-off. Tell me which of those two you would rather be having, and why."
"Think about the handover you run every Monday. What is the one thing that, if it went wrong, would be hardest to recover from that week?"
Illustrative. The wording of any given turn is generated live. The mechanism is a standing rule of the Zavmo Teaching Charter. Every concept is framed as "in your role as X, this means". Every example is drawn from the learner's own stated industry. There are no exceptions.
The personalization of examples by connecting them to one of my hobbies. That made the content feel much more relevant and engaging.
University researcher, expert testerWritten feedback on the learning experience, August 2026Initially, some of the learning section titles felt somewhat disconnected from what I had shared during onboarding. However, once I entered the activities, I realized they were in fact more personalized than they initially appeared.
Entrepreneurship researcher, expert testerWritten reflections on the learning experience, August 2026Eight things Zavmo knows, and what each one changes.
Behind every turn sits a bundle of around fifty fields, the single source of truth for what Zavmo knows about a learner at that moment. It is gathered as the learner goes and folded into the brief of all four stage tutors, every time they speak. Worth being specific about, because vagueness here is how the industry gets away with calling three pre-made tracks adaptive.
| What it knows | What that changes in the lesson |
|---|---|
| Who you are | Your preferred name is used. Prose density and chunk size are calibrated for a stated neurodivergent profile, silently, without ever naming it back to you. |
| Your work | Scenarios reference your actual world: your role, employer, industry and what your days contain, rather than a generic case study. |
| Your education | The tutor knows whether you are returning after years away, and how that last attempt landed. Coming back after a bad experience is treated as a fact about you, not a deficit. |
| Your direction | Lessons are framed against the goal. This is the conversation people in your target role have. |
| Your inner life | Where you have said your confidence wobbles, the tutor goes warmer. Where you named something you are proud of, it anchors there. |
| Your sensibility | Favourite books, films and music become a metaphor library the tutor can reach into when a topic suits an analogy. |
| Your practical fit | Pace, session length and preferred explanation style drive the rhetorical structure, whether story-led, first-principles, step-by-step or visual, and steer away from what you flagged as boring. |
| Your skill check | Your self-rating and diagnostic answers anchor the scaffolding to where you actually placed yourself. |
Two details matter more than the list. First, assessment deliberately uses a leaner mix, identity and current work and inner life only, because proving what you know is about what you know now, not where you are heading. Your ambitions do not get a vote in your own marking. Second, if the bundle ever fails to build, the tutor falls back to a thin profile and teaches anyway. The lesson never fails because the personalisation did.
The ways it adapts, and what each one feels like.
These are not stages in a sequence. They run at once, on every lesson. Each is labelled by the thing it adapts to, and marked live or rolling out, because built and switched on are different things.
It checks your answer about yourself
Before teaching begins, a short Bloom-graded diagnostic asks you to place yourself on each skill, then asks for evidence of the level you claimed. If the evidence does not support the claim, it adjusts.
Zavmo keeps both numbers separately: the level you said, and the level your evidence supports. Modesty does not get you taught things you already know. Bravado does not get you dropped in over your head.
It teaches one rung above you, deliberately
Say your validated level on a skill is Apply. Zavmo then builds that part of your qualification from units at the level above. Every lesson is new ground, not revision dressed up as a refresher.
That removes the most demoralising feature of corporate training, sitting through what you already knew, and its twin, being handed something four levels beyond you and quietly concluding you are the problem.
Every concept lands in your world
Contextualisation is a charter rule, not a feature. Every concept is framed as "in your role as X, this means", every example comes from your stated industry, and the relevance is tied explicitly to the role you are working towards.
The tedious translation step disappears. On a conventional course you learn the abstraction and then, on your own time, work out what it means for your job. Most people never do the second part.
Twelve teachers, not one
The Builder lays foundations. The Coach gets you applying it to your real work. The Navigator has you weighing judgements. The Explorer has you designing something new. A specialist panel brings several perspectives to one complex case, then asks you to teach it back.
One intelligence plays the whole cast, handing over with a note each time, so you get twelve voices and one coherent teacher.
It notices you are struggling before you say so
Agent 13 reads your flow state from the live signals of the conversation: how long you take to reply, how much you write, the confusion in what you say. Most of the time it observes and does nothing. When flow drops it first adapts the current tutor, then switches only if that fails.
The restraint matters as much as the intervention. It never interrupts peak flow, and where a peak holds it moves you up to a steeper tutor rather than holding you back.
Neurodiversity is the architecture, and it is never mentioned
Three teaching characters exist because ADHD, dyslexic and autistic cognition are strengths worth building a teacher around. Agent 13's flow reading is itself neurodiversity-aware.
The charter then makes one rule absolute. The platform celebrates different ways of thinking but never references a learner's neurodivergent profile back to them. It adapts silently. For anyone who has spent a career being accommodated loudly, that is the whole point.
A wrong answer costs you nothing
Teach the concept with a concrete example from your world, check with a single open question, then evaluate. Partly right gets a clarification and a re-ask from a different angle. Weak understanding gets the concept re-taught with a different analogy, not the same explanation more slowly.
Progression is mastery-based, so a wrong answer carries no penalty at all. This is the mechanism that most changes how learning feels.
Forgetting is designed for, not fought
A daily durability check scans your record and, where a competency has not been demonstrated for around three weeks, raises a spaced re-check. The idea comes back at roughly the point it is about to fade, rather than at the point you happen to log in.
Built and working. The standing scheduled job is still rolling out rather than on for everyone, which is why it is marked amber here.
Your learning becomes something you can look at
In the Mind Palace every concept you have touched is a point of light, brighter and larger the higher the level of thinking you reached on it. Connections age from forming through growing and stable to mature. You do not build it. Your learning builds it, from the evidence record.
The Learning Diary is Zavmo's written account of each lesson, drawn from what actually happened. If a lesson produced too little evidence to write about honestly, Zavmo writes nothing at all rather than inventing a reflection.
It ends in something the world already recognises
Zavmo does not invent content and award a badge of its own. It assembles a real qualification from the UK regulated register around your specific gaps, and shows you the whole structure for approval before a single lesson begins.
Every step is traceable from your first answer to the certificate. Completion is explicitly not the standard. Demonstration is.
Speak it, or read it, or have it read to you
A learner can answer out loud instead of typing, using the microphone beside the chat box, and any passage can be played back with "Read it to me". Zavmo also learns a modality preference over time from the learner's own resonance ratings.
Think of a dyslexic learner, or someone working in their second language, or anyone who thinks better out loud. The channel stops deciding whether they can show what they know.
You can speak directly to the system instead of typing a response. Simply click the "microphone" icon next to the chat box. I find this method (speaking instead of typing) particularly helpful.
Programme director, university partnerBriefing the expert testers before the pilot, July 2026The conversational style of the AI. I really appreciated how it responds with kindness when I make mistakes and gently redirects me when I try to skip a question instead of simply blocking me.
University researcher, expert testerWritten feedback on the learning experience, August 2026Four doors, in order.
The eleven mechanisms run at once. The journey they run inside is genuinely sequential, and a stage cannot open until the one before it passes its check.
Discover, find the gap
Your role, your goal in your own words, the target role, and the specific standards sitting between the two.
Discuss, build the qualification
The diagnostic establishes your validated level. Zavmo assembles a real regulated qualification one level above it, and you approve it before teaching starts.
Deliver, teach it
Each learning outcome becomes a conversation, climbing from understanding to application to judgement to creation, with a different tutor at each rung.
Demonstrate, prove it
Verification questions test whether you can integrate and apply it. A gap reopens only the specific lesson concerned, with no penalty. Then, and only then, the certificate.
The way the learning journey is structured into sections such as Understand, Diagnose, etc. It creates a clear progression.
The instructional design, with learning objectives, learning outcomes, and assessments clearly articulated and well aligned with course design standards.
University researcher, expert testerWritten feedback on the learning experience, August 2026What actually changes for the person being taught.
Strip out the architecture and four things change about the experience of learning. These are what the testers reached for when asked what was different, rather than any feature.
- You are not asked to pretend. The system checks where you are rather than taking your word for it, teaches above that point, and treats a wrong answer as information. There is nothing to bluff and nothing to be caught out on.
- You never have to translate. The examples arrive already in your world. The step where you privately work out what any of this has to do with your job has been done for you, on every concept.
- You are not managed into a category. The adaptation to how your brain works happens silently. Nobody labels you and nobody accommodates you in front of the class.
- You leave holding something. A recognised qualification, a map of your own knowledge, an honest written account of every session, and the underlying record. Downloadable, disputable, yours.
He described the concept, flow, and direction as excellent and viewed the experience as a genuine transformation journey rather than a conventional online course.
Entrepreneurship professor and accelerator leaderTelephone feedback, written up by the programme director, August 2026Against a conventional course platform
The distinction he drew was between content that exists before you arrive, and a journey that does not exist until you do.
He believed the platform represents how learning should be delivered because it adapts to the individual user rather than placing everyone into the same standardized learning experience.
Entrepreneurship professor and accelerator leaderInitial user feedback, as recorded by the programme director, August 2026In his view, traditional platforms provide predefined content and videos that users complete in sequence, while [Zavmo] creates an AI-driven journey tailored to the individual.
Entrepreneurship professor and accelerator leaderOn the comparison with conventional course platforms, August 2026Against a general-purpose AI chatbot
This is the comparison that matters most commercially, and the one the market least understands. A chatbot will answer anything you ask and remember roughly where you were. It has no standard to hold you to, no view on what you should learn next, and no obligation to notice you have drifted.
He observed that general AI conversations can be random, unstructured, and prone to losing the focus or context of an ongoing discussion. In contrast, he believed the [Zavmo] experience was designed around a defined transformation pathway and could retain focus on the learner's intended journey.
Entrepreneurship professor and accelerator leaderOn the comparison with general-purpose AI chatbots, August 2026What personalisation is actually for
Not engagement. Movement. The same tester reported real work done on his own thinking, inside a single early session. It helped him separate innovation from transformation, against the venture he was building. That is what a personalised lesson buys you that a generic one cannot. It can operate on the thing you actually brought with you.
He appreciated that the platform continually asks what the user has done and connects those experiences to the learning content, and adjusts the journey rather than treating every learner in the same way.
Entrepreneurship professor and accelerator leaderInitial user feedback, as recorded by the programme director, August 2026What impressed me most is its ambition to move beyond content delivery toward personalized, action-oriented learning.
Entrepreneurship researcher, expert testerWritten reflections, August 2026. She went on to ask whether a research paper was planned on the modelA course you finish. A companion you come back to.
There is a second-order effect in the pilot feedback that is easy to miss. A conventional course has a natural end. You complete it and you leave, and nothing you built inside it belongs to you afterwards. A system that has learned who you are is in a different relationship with you. It holds your history. It can show you the map of what you know. That is a different reason to come back.
Because the system learns about the user and retains information about their journey, he felt users would have a reason to return repeatedly rather than complete a course and leave. He also believed the accumulated personal learning history could create meaningful continuity.
Entrepreneurship professor and accelerator leaderInitial user feedback, as recorded by the programme director, August 2026
The words he used for the mechanism were emotional connection. Unlike a course you complete and abandon, this one carries your history, goals, context and direction with it. Whether that holds at scale is exactly what the wider pilot exists to find out. It is the right thing to be testing, and it only becomes testable once personalisation is real rather than cosmetic.
Built, and switched on, are different things.
A page about personalisation is worth very little if it quietly claims capabilities that are not in front of learners today. So, plainly.
The full four-stage journey. The twelve-tutor model and the five-rung teaching flow. Agent 13 flow orchestration. The competency graph, gap analysis, teach-one-level-up qualification build and certification. The fifty-field personalisation bundle, wired into all four tutors. The Mind Palace. The Learning Diary and journal, with download. The xAPI evidence layer.
Cross-session memory, the route to Zavmo genuinely remembering a learner between sessions, is built, privacy-first, and currently shipped dark and off by default. On-demand translation and in-language tutoring, the standing spaced-durability job, and the fuller server-side Agent 13 are in the same category.
Video-led lessons, avatar coaching and a continuous two-way spoken conversation were asked for by the testers, and those are roadmap. Voice itself is not: learners can already answer by microphone and have any passage read aloud. What does not exist yet is a natural spoken dialogue with a visible coach.
The testers had real criticism. Most of it landed on the onboarding, where the purpose and length of the discovery questions were not made clear enough. The rest landed on the interface, where the central chat area was judged too cramped. One tester waited a long time for a learning journey to generate. Those are execution problems, they are being fixed, and this page is about the learning itself.
What matters is that when the same testers were asked what worked, they did not describe features. They described being taught.
Bloom asked the question in 1984.
A class of one, for everyone, at the cost of a software interaction. Come and see it working.
Book a demo