Organisation type
Civil Society Operations
A learning programme on programme design, proposals, monitoring, evidence, reporting, communication, due diligence, safeguarding and privacy in civil society organisations.
27 courses · 2 specific to this profession
Courses for this type of organisation
Courses tied directly to this work appear first.
Courses that apply to every profession
The foundations of working with AI, checking output, agents and getting unstuck. The same for every field.
Microcourse (18)
Estimated study time: 5 hours
Agents: Installing one and finding the stop button
An agent acts, it does not only write. Before it acts on your laptop: the source, the dialog, the first run in a safe place, and the stop button.
Estimated study time: 5 hours
Agents: Permissions and scopes
A scope is a sentence: who, what, where, how long. Least privilege, the permission screen, keys and tokens, and revoking as a habit.
Estimated study time: 5 hours
Agents: What an agent may touch
The scope says what it may do; the boundary says where. The allow list, the test copy, the data levels for agents, and the dry run before any real run.
Estimated study time: 5 hours
Agents: Reading the action log
The output tells you what the agent produced; the log tells you what it did. The ten-minute review after every run, the three signals, and what a missing log means.
Estimated study time: 5 hours
Agents: Undoing an action
A read cannot be taken back, a write reverts with a version, a send cannot be unsent, a run depends. The copy before, the rollback after, and the incident when there is no undo.
Estimated study time: 4 hours
The four part request
Role, context, limits and output form: learn to structure a request for a routine text task and check the tool's response.
Estimated study time: 4 hours
Checking an AI output
Facts, tone and organisational rules: three short checks after each AI output to look for common errors.
Estimated study time: 5 hours
Fabrication: The four kinds and how to catch them
Numbers, citations, authorship, dates: the four fabrication types, the three early signs, and the ninety-second citation check.
Estimated study time: 5 hours
What you may paste: Data levels and the law
Four data levels, two laws, a ten-second test before every paste, and what you do when the data have already gone out.
Estimated study time: 4 hours
Disclosing AI assistance
One line that says an assistant helped, where it goes, the record that backs it, and how it becomes a team habit.
Estimated study time: 5 hours
Trigger, action and connector
A flow is a trigger, some actions and the connectors between them. Before any tool builds it, you draw it on paper and read it part by part.
Estimated study time: 5 hours
Credential, retry and log
A flow fails at three in the morning. The credential that expired, the retry that sent five times, the log that did not exist. Three decisions before the flow is switched on.
Estimated study time: 5 hours
The human check before anything is sent
A person before every send and every run. Where the check sits, the screen the person sees, the time they get, and what happens when nobody answers.
Estimated study time: 5 hours
A flow broken five ways
One flow, five breakages, one method: from the symptom to the part. No trigger, wrong connector permission, expired credential, a loop running twice, output to the wrong place.
Estimated study time: 5 hours
Reading the stop
When a tool stops, the message is information. Learn to copy it exactly, to note what you did just before, and to ask what it means before you ask how to fix it.
Estimated study time: 5 hours
The smallest step, then the check
Never three commands in a row. Learn to do one step, check it, and undo it before you do the next.
Estimated study time: 5 hours
Permissions and the company laptop
Read a permission dialog, refuse what must not be granted, ask your company's technical support for exactly the right thing, and know when a web version replaces a local install.
Estimated study time: 5 hours
When to stop
The signs that show a problem will not yield to more effort, and a four-question decision ladder that tells you where to stop and what to change.
Training (7)
Estimated study time: 9 hours
AI foundations: Working with ChatGPT, Claude and Gemini
Comparative tool literacy: which AI assistant for which task, and why, based on how each one actually behaves rather than on marketing.
Estimated study time: 9 hours
AI foundations: Ethics and risk
Fabricated answers, confidentiality, source verification, bias. Six sessions on the real risks of AI in professional work and how to manage them.
Estimated study time: 12 hours
AI foundations: AI and the Albanian language
What works, what breaks, and how to compensate. Six sessions on how AI behaves with Albanian: dialect, code-switching and technical terminology.
Estimated study time: 12 hours
AI foundations: Getting started
What AI is, what it is not, and how to think about language models without mystification. Six foundation sessions for any working professional who wants practical AI knowledge.
Estimated study time: 12 hours
AI foundations: Prompting in practice
Prompt structure, role prompting, structured output, iteration. Six practical sessions that change the way you write with AI.
Estimated study time: 9 hours
AI foundations: Evaluating AI output
How to judge whether an AI output is good. The evaluation mindset, criteria, and the habit of a second review. A bridge session towards professional evaluation work.
Estimated study time: 6 hours
The environment card
Operating system, rights, terminal and company rules: one paragraph that makes every question you ask easier to answer.