A decision map for AI, built from capability categories and risk — not from chasing model names or tool trends.
This track builds a decision map for AI by teaching capability categories and risk, instead of chasing model names or the latest tool trend. By the end, a learner can explain the tool categories, choose a method based on evidence, data sensitivity, and risk, evaluate a vendor's feature claims critically, and propose a governed experiment rather than an unmanaged rollout.
The approach P1 uses throughout: from remembering tool names → to understanding capability categories → to choosing by task and data → to evaluating claims → to experimenting with governance.
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.
Know the main categories of AI — general assistant, knowledge-grounded, persistent assistant, research, creative, productivity, automation/agent, company system — and use the map instead of a list of apps.
Understand well enough how systems generate answers, and why they can be confidently wrong — probabilistic output, context, hallucination, tool use, model updates.
Choose ChatGPT, Gemini, Claude, or an equivalent by task, context, plan, and policy — never by model name alone.
See the difference between a throwaway chat and a workspace with custom instructions, knowledge, and version control.
Choose a source-grounded tool when evidence matters — source-grounded answers, inline citation, source selection, coverage gaps.
Know the difference between the model's built-in knowledge, web search, and deep research — recency, primary source, the citation trail.
Decide when creative media genuinely helps the work and when it should not be used — a creative brief, rights, likeness, disclosure, a fact check.
Understand AI embedded in documents, email, meetings, and spreadsheets — in-flow productivity, organisational data, permissions, enterprise data protection.
Understand levels of automation — retrieve, draft, recommend, act — and why they must start narrow: an approval gate, logs, least privilege.
Assess data before using a tool, and know the escalation point — classification, retention, access, approval, auditability.
Evaluate tools and features by real task, evidence, and total cost — use-case first, a pilot, a rubric, adoption, lock-in, change management.
Turn the team's AI map into a portfolio that can actually be developed — standardise, sandbox, or retire, with an owner and a review cadence.
This track does not teach a list of apps. It teaches capability categories and selection criteria, because app lists expire and criteria do not. The prompt below is week 11's criteria in a form you can use on whatever proposal is on your desk right now.
This track is deliberately vendor-neutral. Week 3 teaches choosing between ChatGPT, Gemini and Claude by the nature of the task, the context, the service plan and company policy — never by brand.
To keep the names straight: a persistent assistant is called Gems in Gemini and Projects in Claude and ChatGPT. NotebookLM belongs to a different category altogether — answering from a defined set of sources.
Features and limits shift with plan, region and account. Open the official documentation linked at the foot of this page before concluding what any tool can do.
Context: someone has proposed that my team start using a new AI tool, and I do not want to decide by hype. Task: help me evaluate it in five steps 1) sort my team's main tasks (listed below) into capability categories, not app names 2) say which category this tool sits in, and where it overlaps what we already have 3) separate what the vendor's official documentation states from what the marketing claims 4) raise the data, access-permission and approval questions we must answer before any trial 5) propose one small reversible experiment, with success and stop criteria set in advance Constraints: where you are unsure about this tool, say you are unsure and point me to the official documentation page I should read. Do not guess pricing or features. Format: steps 1-2 as tables, steps 3-5 as lists. Tool proposed: [tool name] My team's main tasks: [list 3-5]
Let the tool help you evaluate, but an AI answer is neither evidence nor an approval. Any real trial still goes through whoever owns the data and access decision under company policy.
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.
This track works well as an orientation before Track 01, or as a governance layer once a team is already using AI day to day.
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