Build basic, working fluency with AI: safe use, checkable prompts, verified sources, and helpers you actually reuse.
This track builds basic fluency in using AI for real work — safely, verifiably, and without producing shallow output. By the end, a learner chooses the right tool and method for the task in front of them, produces a draft that can actually be used, checks its facts, and delivers the work with evidence attached: what went in, what was assumed, how it was checked, and who owns it.
The approach P1 uses throughout: from a one-off instruction → to a task with context and criteria → to sourced evidence → to a verified, usable result.
Every session follows the same arc: connect to real work → teach the concept → demonstrate it with and without context/evidence → practise it → apply it to one real, approved task → reflect and set the risk-checked homework.
Separate what AI can help with, what needs a human, and what data must never be entered — without giving up ownership of the decision.
Write prompts with a goal, context, data, constraints, format, and quality bar — so the output can be checked and revised.
Use AI to draft documents while keeping the owner's voice and accountability: structure before style, and always check names, numbers, dates.
Turn long text into summaries, decisions, and follow-up lists without inventing facts — and name what wasn't found.
Use web search with evidence, and separate the primary source from opinion. AI can fabricate citations.
Design a Gem around persona–task–context–format that reduces repetitive typing without doing the thinking for the person.
Package knowledge, instructions, templates, and a decision log so AI understands the work continuously — updatable, and owned.
Build a notebook that answers only from a defined source set, with citations you actually check — citation is not the same as truth.
Use NotebookLM/AI to turn sources into a briefing, FAQ, checklist, or training material — matched to the audience, checked before release.
Understand the sequence — ingest, clean, structure, check, reuse — with metadata, schema, provenance, an owner, and a review gate.
Create images, slides, or media only when they genuinely help communication — with a brief, an audience, and a rights and fact check. An image is not evidence.
Combine the twelve weeks into one real, repeatable workflow: human review built in, and only as much prompt and evidence kept as is actually needed.
The whole track reduces to one habit: give an instruction complete enough that the result can be checked. Below is a real prompt built on the same frame taught in weeks 2 and 4. Copy it as it is.
This prompt works in any general assistant — Gemini, Claude or ChatGPT — because it is instruction only and leans on no vendor-specific feature.
The difference appears when you want to save it for reuse, because the feature names differ. Week 6 teaches Gems, which is Gemini's name and only Gemini's. The equivalent in Claude and ChatGPT is called Projects. Week 8 uses NotebookLM, a separate Google product rather than a mode inside Gemini.
Following one vendor's instructions on another vendor's screen will simply fail, and that is not your mistake. Check the feature name against the screen actually in front of you first.
Context: I own this piece of work, and I need to summarise the material below for my manager today. Input: the text I have pasted at the end of this prompt is an unstructured email thread or meeting note. Task: give me three parts 1) the facts explicitly stated in the source 2) the decisions already made, each with its owner 3) the follow-up items, with deadlines where the source states them Constraints: do not guess and do not add anything absent from the source. Where something is missing, write "not found in source". Format: a concise table, no longer than one page. Quality bar: finish with a heading "What I must check myself" listing three items — pick the points where being wrong would cost the most. Text: [paste your email thread or meeting note here]
Before pasting, replace names, customer identities, financial figures and personal data with training or masked values, and use only a tool your company has approved. This is the same safety line the course applies every week.
Facilitation rule throughout: use only anonymised or constructed P1 examples; pair learners as task-owner and checker; measure decision quality and reusability, never the number of prompts or apps tried; and if a use case touches sensitive data, finance, personnel, or system access, stop and escalate through P1's approval channel.
Instructor note: open the official references below before teaching the related session, and match button names and plan tiers to whatever P1 has approved on the teaching day.
Once a team has this foundation, the natural next step is learning to think with AI rather than only instruct it.
This is the curriculum P1 Thailand built and uses to train its own team, before teaching it to anyone else.
Twelve weeks, one shared foundation, applied to each person's real work.
Request a consultation