One-to-one distance learning · First cohort, January–June 2027
One participant, one instructor, one taught hour a week and two support slots availible, building systems around the work you already have to produce.
Conversation is thirty minutes, no cost, about your work rather than the course. Or download the syllabus first.
This is the pilot session of the Practicum. January 2027 is the pilot cohort. Instead of testimonials, see our past work in action..
The shape of the twenty-two weeks
Four required artifacts, two working tools you built, and the Playbook — handed over complete at the end ouf the course.
The shape of it
Twenty taught sessions across twenty-two weeks: three units of six sessions each, then two sessions of consolidation. Two of those weeks are scheduled breaks, agreed in advance with flexibility if needed.
Beside the taught hour, two support slots are held open every week. These are fifteen to sixty minutes each, used as the week needs them, included in tuition. Designed for unsticking stuck homework and fixing failing builds together.
Each unit closes on a required piece of work. If the work is not there, the unit is not finished. The fourth unit is not a fourth subject, it is where the first three are consolidated into something you keep.
The 22 weeks
Who it is for
Three seats per cohort, and the fit matters more than the fee. If you are not one of these, the program may be the wrong fit for you. Lets talk on a call and see if this would be helpful in your work.
You produce briefs, memos and background research for clients. Your ceiling is not demand. It is how many hours of reading and synthesis you can personally do in a week.
Your subject matter punishes sloppiness. You need to know where these AI tools break before you will put your name on anything they touched.
You are working remotely from the market you built your career in, where institutional research support is more difficult to come by, and you are competing against those with local resources.
This is not a coding course, a machine-learning course, or a university certification. It is a practicum. You leave with working systems and tools you built, along with a certificate stating your completion of these practical accomplishments.
Outcomes
Seven of them. Each one is demonstrated by something you produced, not by a session you attended.
Build and maintain a grounded research corpus that accumulates across engagements, so that reading, interviews and analysis stay retrievable a year later instead of being lost when a project closes.
You leave with: a live research notebook system and a written architecture standard for it.
Use AI to tnterrogate a body of sources critically — extracting structured comparisons, testing a claim against the evidence behind it, and constructing the strongest version of a position you disagree with.
You leave with: a documented prompt library for analytical work.
Verify AI-generated claims, and diagnose a failure by naming which property of the system caused it, so that work with AI in the process is still work you can defend.
You leave with: a verification checklist and a failure-diagnosis sheet, applied to your own output.
Run recurring administrative work — documents, correspondence, commitments, scheduling — through configured, persistent systems rather than starting from nothing each time.
You leave with: configured workspaces, two working skills, and a documented workflow for each recurring task.
Write a specification precise enough that a working tool can be built from it, which is the actual skill behind turning “I wish something did this” into something that exists.
You leave with: an intent file and specification, and the tool built from them.
Build, test, host and hand over a small single-purpose web tool, then do it a second time without help.
You leave with: two working tools, live at URLs, with handover notes.
Set and defend your own limits on AI use: what never goes into a model, what has to be disclosed, and the terms on which you let AI-assisted work leave your desk.
You leave with: a written personal AI use policy, in three parts.
What is included
Roughly US$25–60 per month on tool subscriptions during the course, depending on the plans you choose. That is paid directly to the providers.
Structure
Three units of six sessions, then two of consolidation. Two of the twenty-two weeks are scheduled breaks, which is honest about what five months of anyone's calendar looks like.
The unit that pays for the course. Consulting research is expensive because reading and synthesis are slow, and because almost nothing survives the end of an engagement. This unit builds a system where the reading accumulates instead. It starts with how these tools fail, because everything after that depends on knowing where the edges are, and it ends with you producing a real research brief under a procedure you wrote.
Administrative load is the tax on independent work. Persistent workspaces, recurring documents built as skills you own rather than prompts you paste, correspondence, a commitments system that you will actually keep using, and a clear-eyed session on what should not be automated at all. Ends with the confidentiality question that regulated work turns on: what never goes into a third-party model.
Where you stop being only a user. Not programming, but instead development and review through specification, direction, testing and hosting. One tool you need is taken from an intent file and written spec to a working, hosted, documented thing over four sessions. Then a second, of your own choosing, built inside one session, to establish that the method transfers and does not depend on me being there.
Portfolio review against all seven outcomes, gaps closed, and a written twelve-month plan for what to maintain, what will decay, and what to learn next. The Playbook is handed over complete.
In the mark, this is the solid letter — the three bars resolving into something you can point at. It is two sessions, not a fourth subject.
Published in full, and fixed in structure. What moves is the material inside a session — shaped by what our conversations show you actually need, by what these tools can do by the time we get there, and by whatever I have learned since about teaching the hour better. What is fixed and what moves.
Method
If the hour produces nothing you can point at, the hour failed.
That is the whole method, and it is why the course is one taught hour a week for five months rather than a weekend intensive. Skills you use once do not survive; systems you built and kept using do.
I do not lecture for an hour. Each session has one thing to build, and we build it on your material while you drive. You keep the file, the notebook, the prompt or the tool.
That constraint shapes the whole curriculum. It is why the calendar tool in Unit 3 is a tool you actually need rather than a teaching example, and why Unit 1's sources are drawn from work you are being paid for. It is slower than a syllabus, and it is what the informal version was missing. The systems I am still using three years on are the ones that came out of a real piece of work, not the ones I was shown once.
If you would rather see that than read about it, there is a finished tool on this site — the kind Unit 3 builds, with the written account of the four sessions it took. Open it and paste something in. The rest of what these methods have produced, with the counts, is on the instructor page.
You need seventeen of the twenty main sessions and all four unit artifacts. Support sessions are optional and are counted neither way. If the artifacts are not there, the course is not complete, and the certificate does not go out.
Evidence
January 2027 is the first cohort. This season will host the first classes of this practicum. To review for your own confidence: the methods this course teaches have been running a business in front of paying customers since April 2023, and those customers left a public record of how it went.
“the planning that went into every moment of our trip was unlike anything I have seen; from the meals to plan A, B, and C riding options to road stops for tea; this team is thinking several steps ahead at all times.”
“We had some issues outside of the control of A.A with permits, and they pivoted our itinerary very quickly making seamless changes to the itinerary as necessary.”
“Communication and packing lists were clear and easy to follow.”
Three of 21. All of them, including the one that is not five stars, are on the public review page, in full and in their own words rather than in mine.
Itineraries that held. Documents that were right. Plans that had a B and a C in them, and a permit failure absorbed without the trip falling over. The systems taught in this course are what those things run on — the document generators, the trip sites, the research and the contingency work behind an itinerary, built and kept the way Units 1 to 3 teach them. The reviewers are describing the output. None of them was asked about AI and most have no idea it was involved, which is the part that makes it worth reading: the work either stands up in front of a paying customer or it does not.
Instructor
I am not an AI company. I run Adventure Access in India, an adventure travel operation working in Ladakh and Zanskar in the Indian Himalaya, plus the rental gear business that supplies it. Itinerary design for mixed-nationality groups, air cargo into Leh, Indian regulatory filing, and a large volume of client-facing documents that all have to be right.
Since 2023 I have rebuilt most of how that gets done around these tools. Documents that took an afternoon take twenty minutes. I have built small tools for myself that I could not have commissioned at any price making sense for a business this size. People started asking me to show them, the informal version did not work, so I built a course.
That is a large claim to make about yourself, so the instructor page carries the record behind it: eighteen client trip sites built and deployed, a guest document intake system, a receipts and reconciliation pipeline seventeen months deep, generators for Indian foreigner registration, and the tools on this site. I trained as a mechanical engineer, programming coursework included. I have never worked as a developer, but I can read code well enough to tell when it is wrong, which is the skill Unit 3 actually teaches.
I am an experienced practitioner, not a researcher and not a data scientist. Where the course reaches the edge of what I know, I say so.
Tuition
$3,300 covers twenty taught hours of one-to-one instruction — $165 per main session — plus up to two support sessions every week, curriculum design, written materials, the tool-build support in Unit 3, assessment and certification. Used at all regularly, the support slots put the real cost of an hour of my time well below that headline figure.
For comparison: private one-to-one professional coaching averages around $234 an hour according to the 2025 ICF global study, and university executive-education AI programs run $190–380 per contact hour — delivered to a room of thirty people rather than to one. The Practicum is priced below both because it carries no institutional overhead, not because it is a lighter program.
Nothing is invoiced until after the third session, and the schedule after that is agreed with you. Participants funding tuition through an employer or a professional development fellowship should read the funding notes.
Complete program
$3,300
20 taught hours over 22 weeks, one to one, plus weekly support.
Nothing is billed until after the third session. If it is not what you need by then, say so and you owe nothing.
Questions
You would be in the pilot cohort, and I would rather say so here than have you work it out in March. What that actually means: the curriculum is written and published in full, the syllabus is downloadable, the methods are ones I have been developing since April 2023 on work with real consequences, and the first three sessions are unbilled — so you watch the hours run before any money changes hands. What it does not mean is that I can show you a graduate. Nobody has finished this course, because nobody has started it. If that is the difference between yes and no for you, it should be no, and I will not try to talk you round.
Not for the course, and I will not manufacture any. What exists is 21 public reviews of the travel business these methods were built to run, rated 4.5 out of 5, collected by Trustpilot rather than by me. The ones that speak to the planning, the documents and the communication those systems produce are quoted further up this page, with a link to the rest. They are evidence that the methods hold up in front of paying customers. They are not evidence that anyone has completed this course, and they are not offered as it.
Both, and the line between them is worth being precise about. Fixed: twenty sessions, four units, the required artifacts, the seven outcomes, the completion requirements and the fee. Those are what the certificate records and what a funding body is shown, and they do not move once you have enrolled. Not fixed: what happens inside a given hour. Three things change it — what you tell me you actually need, which is the whole reason the course is one to one; what these tools can do by the time we reach that session, since they are not mine and they change every few months; and my own view of how to teach something, which I would rather act on than leave a worse hour in place out of tidiness. If a change would affect an outcome or an artifact you hear about it first, and you can hold me to the published version.
No. Unit 3 teaches you to specify, direct, test and host small tools, which is a different skill from writing code. You will read code during the course and you will understand what you are reading. You will not be asked to write it from scratch.
One hour of live instruction per week, plus one to two hours of assigned practice between sessions. Call it two to three hours a week for twenty-two weeks, with two scheduled breaks. Around sixty hours in total. The two support slots each week sit on top of that and are optional — some weeks you will use both, most weeks you will use one or none.
Missed sessions can be rescheduled within the program window, subject to both calendars. Completion requires seventeen of the twenty main sessions, so there is room for three unrecoverable absences. Beyond that the course is recorded as incomplete, and I would rather tell you that in June than pretend otherwise. A missed support session is not an absence; those are yours to use or not.
Principally Claude and NotebookLM, because those are the two I use daily and can teach from experience rather than from documentation. The course also covers static web hosting for the tools you build. I use ChatGPT, Gemini and Perplexity routinely in my own work as well, and where a different tool is genuinely better for something you need we will use it and I will tell you what I do and do not know about it.
No. You buy your own, directly from the providers, on whatever plan suits you. Budget roughly US$25–60 a month during the course. Bundling software into tuition would inflate the fee and lock you into my choices.
A Certificate of Completion for The Applied AI Practicum, issued and signed by me as the instructor, recording the program title, delivery dates, contact hours, units completed and artifacts produced. It is a record of completed private coursework. It is not an accredited qualification and it carries no academic credit, and you should not present it as either.
Yes, and this is common. On request I provide an itemised invoice, a course description with hours and curriculum, an attendance record and a completion certificate — the four documents a claim normally asks for. Note that most schemes will not reimburse software or subscription costs, which is one reason tuition and tool access are kept separate here. See the funding notes.
Two support slots every week, fifteen to sixty minutes each, live and one to one, included in tuition and booked by you when you want them. They are for homework help and for developing the thing you are building further than the taught hour had room for. Plus email, with a reply inside two business days. None of it covers me doing the work — if a piece of research or a document needs producing, that is a separate engagement and I will say so rather than quietly absorbing it.
Nothing is invoiced until after the third session, so there is no refund to claim: you say so, we stop, and you owe nothing. Once invoicing has begun the fee stands, because by then the curriculum has been rebuilt around you and a place has been held that someone else could have taken. I would rather lose the fee in January than have you spend five months on the wrong course.
Because the curriculum is rebuilt for each participant, the support slots are real hours held open every week, and the tool-building in Unit 3 is genuinely time-consuming to support. Four would mean doing it worse.
Live, every session, one to one. Nothing is pre-recorded. The main session is a fixed weekly slot agreed at the start; the two support slots are agreed at the same time and booked week by week. Time zones are worked around rather than imposed.
Me, in both cases. The Applied AI Practicum is taught and run directly by Trevor Richtenburg — one instructor, three places, no company in between, and the invoice comes from me. The methods came out of running an adventure travel operation in the Indian Himalaya; that background, the record of what has been built with them, and what I do not teach are on the instructor page.
Everything a buyer or a funding body asks for, published rather than promised on a call.
A question not answered here gets a straight answer by email within two business days. Asking costs nothing and does not put you on a list.
Enrolment
It starts with a thirty-minute conversation about your work, at no cost and with nothing to prepare.
If the fit is wrong I will tell you on that call, and I may be able to point you to something better. If you would rather email: trevor@zanskariadventures.com