Course 1
AI Foundations for Working Canadians
4 weeksEveningsOnlineBeginner
Who it is for. People who have tried a public chatbot, felt both impressed and uneasy, and want a method they can defend at work. No technical background is required. Curiosity and a real weekly task are.
What we cover:
- What a model is doing when it writes. Pattern completion, not understanding — and why that still helps in a briefing note if you stay in charge of the facts.
- Where models fail in ordinary Canadian work. Invented citations, confident wrong numbers, outdated policy language, and fluent nonsense in a second language.
- What you must not paste. Client files, health information, student work, unpublished personnel matters. We practise stripping an example until it is safe.
- A week that has a place for the tool. One recurring task, a written brief for the model, a check routine, and a stop rule when the output is not earning its time.
- How to talk about it with a manager. Language that does not overclaim, and a short note you can share with IT or records staff if they ask.
You leave with: a one-page personal protocol, a de-identified task pack, and a checklist for the next twelve months of tool changes.
Ask about this course
Course 2
Prompt Craft for Documents and Correspondence
2 weeksEvenings or morningsIntermediate
Who it is for. People who already write for a living or as a large part of the job — coordinators, analysts, office managers, communications staff — and who need the draft to sound like their organization, not like a brochure.
What we cover:
- Briefing a model the way you would brief a junior writer. Audience, constraint, source pack, and the sentence you will not allow it to invent.
- Long documents. Chunking a report, keeping defined terms stable, and marking what must be quoted rather than rewritten.
- Correspondence. Tone that matches a complaint, a thank-you, a delay, or a refusal, without leaking extra facts in the helpful-sounding closing.
- Fact checking as a habit. A pass for names, dates, dollar figures, and claims that only look sourced.
- House style. Turning a messy style sheet into instructions a model can follow, and a human can still override.
You leave with: a reusable brief template, a correspondence set in your organization’s voice, and an editing protocol.
Ask about this course
Course 3
AI for Small Business Operations
5 weeksEvening streamApplied
Who it is for. Owners and operations leads in shops where the same people sell, schedule, and follow up. You do not need a data team. You need fewer repeated hours on quotes, product copy, and “where did we put that.”
What we cover:
- Quotes that stay honest. Frames for scope and exclusions; a hard rule that quantities and prices are never invented by a model.
- Client mail. First replies, status notes, and close-outs that sound like your firm and can be sent by more than one person.
- Schedules and reminders. What is safe to automate, and what still needs a human who knows the site or the season.
- Product and service descriptions. Consistency across a catalogue without copying a competitor’s claims.
- A company knowledge base. The ten pages staff actually search, written so a new hire and a model can both use them.
- A stop-the-line list. Tasks that look automatable and are not — refunds, safety, anything that changes a number on a job.
You leave with: an operations playbook, a starter knowledge base, and a mail set your team can share.
Ask about this course
Course 4
Data Sense Before AI
4 weeksEveningsBeginner to intermediate
Who it is for. People who live in spreadsheets and have been told to “just ask the AI.” You want to know which questions belong in a table, which belong in a model, and how to read an answer that sounds finished but is not.
What we cover:
- Shape before sparkle. Columns, missing values, duplicates, and the quiet damage of a merged cell.
- Questions a table can answer. Counts, filters, joins you can explain to a colleague. If you cannot explain it, we do not ask a model to hide it.
- Questions a model should not answer from a dump of rows. Causal claims, forecasts dressed as facts, and summaries that drop the awkward cases.
- Reading output like an auditor. Spotting rounded numbers that never appeared in the source, and totals that do not add.
- A handoff note. How to describe a dataset so the next person — or the next tool — does not misuse it.
You leave with: a data-hygiene checklist, a “table vs. model” decision card, and a worked example from your own (de-identified) sheet.
Ask about this course
Course 5
Automations Without Code
3 weeksEveningsApplied
Who it is for. Coordinators and office leads who are already copying between forms, mail, and sheets, and who want a chain that fails loudly instead of silently.
What we cover:
- The boring map first. Trigger, handoff, human check, and what happens at 4:55 p.m. on a Friday when a field is blank.
- Forms to sheets to mail. A small, maintainable path using tools most organizations already pay for.
- Where an assistant belongs in the chain. Drafting, tagging, or summarizing — not approving, not sending unsupervised to a client.
- Limits. Rate limits, broken logins, a colleague on leave, and why “set and forget” is how errors become policy.
- Service, not sculpture. A one-page runbook: who owns it, how to pause it, how to explain it to IT.
You leave with: one working automation on your stack (or a documented design if your IT must install it), plus a maintenance runbook.
Ask about this course
Course 6
Privacy-Aware AI Practice
2 weeksIntensive eveningsAll levels
Who it is for. Anyone who handles client, student, patient-adjacent, donor, or employee information and is being asked to “try ChatGPT on it.” This is practical training, not legal advice.
What we cover:
- Classification you can use on a Tuesday. Public, internal, confidential, and “never in a consumer chat” — with examples from ordinary Canadian offices.
- What cannot go in. Identifiers, health and education records, unpublished complaints, children’s data, and anything you would not put on a postcard.
- Internal rules that people will follow. Short, specific, posted where the work happens. We draft; your counsel reviews if you have counsel.
- Consent and vendors. What “the tool’s terms” actually mean for a file that is not yours to donate to a training set.
- Logs. A light journal of who used what, for which class of task, so you can answer a reasonable question later.
You leave with: a draft internal AI-use rule, a red-line list, and a logging sheet. This is practical training, not legal advice.
Ask about this course
Course 7
Build a Custom Assistant
6 weeksProject studioIntermediate
Who it is for. People who already have a body of internal writing — how-we-do-things pages, scripts, product notes — and need a colleague-facing assistant that stays inside those rails. You should be ready to test it with someone who was not in the room.
What we cover:
- Scope that can be refused. What the assistant will do, what it will decline, and the sentence it uses when it does not know.
- A knowledge base worth indexing. Cleaning, dating, and cutting pages that contradict each other.
- Instructions as an artefact. System guidance written so a successor can maintain it.
- Tests. A packet of questions, including rude ones, off-policy ones, and “please invent a discount.”
- Handoff. Who owns updates, how to retire a page, and how to brief the rest of the team without theatre.
- What we will not build. Production systems that touch money, identity, or medical advice. Those stay with your specialists.
You leave with: a documented assistant, a test log, and a handoff note your colleagues can run.
Ask about this course
Course 8
AI for Instructors and Internal Trainers
3 weeksEveningsEducation
Who it is for. College instructors, corporate trainers, and people who write onboarding. You need materials, a stance on student or staff work, and language that does not pretend a model is a marker.
What we cover:
- Materials. Outlines, examples, and alt-text that you still verify. Speed is not a substitute for a source.
- Assessment. Tasks that still make sense when a chatbot exists, and oral or process checks that belong in the room.
- Honesty of work. How to talk with learners without a gotcha culture, and how to document your own use so you model the rule.
- Explaining tools. A short in-class demo that shows failure on purpose, because fluency is the risk.
- Accessibility. Transcripts, plain language, and why generated slides still need a human pass for colour contrast and reading order.
You leave with: a course-policy draft, a materials checklist, and a 20-minute teaching script on tool limits.
Ask about this course
Course 9
AI Literacy Intensive for Public and Nonprofit Teams
2 daysWhole-teamIn person or live online
Who it is for. A department that needs a shared vocabulary this quarter, not a six-week staircase. Managers and front-line staff in the same room, with the same red lines.
What we cover:
- A common picture of the tools. What they do, what they invent, what your records policy already implies.
- Worked examples from your queue. De-identified, prepared with you in the two weeks before the intensive.
- A draft internal rule. Short enough to pin beside a monitor. Your legal or privacy office remains the authority; we supply the working draft.
- Role splits. Who may draft with a model, who approves, who never pastes the file.
- A 30-day practice plan. Three recurring tasks, owners, and a review date — so the two days do not evaporate.
You leave with: a shared glossary, a posted rule draft, and a 30-day practice plan for the team.
Ask about this course