Aicoursesforsg

Method

Five steps, one timed task

Aicoursesforsg runs every open class and every private cohort on the same skeleton. We agree the task, we time it, we rebuild the procedure in the room, we time it again, and we hand over the files. The method is short on purpose. Extra slides that are not on the stopwatch get cut.

See the outcome records Book a scoping call

Training desks ready for class

The Fusionopolis room is set for laptops, with a cap of 12 people so that an instructor can watch each learner on the sample. Notes stay on paper and in a shared folder. We do not run the session as a product tour of the assistant. The assistant is the tool that happens to be licensed; the object of the class is the task.

The five steps in full

  1. Step 1. Scoping call (25 min)

    The call covers the name of the task, how often it runs, which assistant is already licensed, and whether an anonymised sample can be used in class. We also ask who will sit in the room: the people who do the task, or observers. Observers are welcome up to the seat cap; they do not go on the measurement sheet unless they also complete the sample.

    Within two working days you receive a written outline: programme name, hours, fee, sample plan, and a yes/no on whether the public demo pack would be used instead of your file. If the answer is the demo pack, we say so in the outline. That is the same limitation as the published records.

  2. Step 2. Baseline measurement

    Before any teaching, each learner completes the sample with their current method. For inbox, that is 20 tickets. For forecast, that is producing the pack including the held-out quarter. For reporting, that is a first complete draft with whatever source discipline they already have. A second person times the work and scores it against a key written before the day.

    Rows are dropped when a learner has never done the task at work, or when a machine failure makes the timing unusable. Eleven clean rows are more useful than twelve patched ones. The baseline sheet is part of the handover; you keep your own numbers.

  3. Step 3. Workshop day

    Instructors rebuild the procedure with the group in the room. The prompt file is written in the open, on the board and in the shared folder, so that nobody leaves with a magic sentence they cannot edit. The checklist is short enough to be used under time pressure. Class discussion is steered back to the sample whenever it drifts toward product features.

    Full-day classes include a 60-minute lunch that is off the clock. Evening hybrid sessions run 3 hours each, with a 10-minute break. Instructors stay after the official end for 15 minutes of desk questions; that time is also off the clock.

  4. Step 4. Build and verify

    Learners apply the new procedure to a held-out slice of the same class of work. For inbox, ten tickets that were not in the morning set. For forecast, the same held-out quarter, now produced with the checklist. For reporting, a second pass on the same pack after the source-marking rule is in place. The second marker uses the original key. We do not rewrite the key to flatter the after-pass.

  5. Step 5. Re-measure and handover pack

    The after timing is taken on the same rules as the baseline. Each learner leaves with four files: the prompt file, the checklist, the measurement sheet with both readings, and a short note on what the second marker counted as an error. A 30-day remote clinic (30 minutes) is booked before people stand up, so that the slot exists while the work is still fresh.

Tools and licences

Aicoursesforsg does not sell software. You attend with a live login to a mainstream commercial assistant that your organisation already licences. Instructors will have a spare demo login for the sample pack if a learner’s workspace refuses the anonymised file; that spare is for the classroom sample only.

We work in the products that Singapore offices actually buy in 2026: the major conversational assistants and the add-ins that sit on top of mail and spreadsheets. We will not spend contact time comparing vendors. If your organisation is mid-migration, say so on the scoping call so that we can keep the prompt file in plain text, portable between the two tools.

Learners should have permission to paste the anonymised sample into the tool. If your security policy forbids that even for stripped data, book an on-site private cohort and keep the file on your network. The method still holds. The public records will not be a match for that constraint, and we will say so in the outline.

What you take home

The handover is the point of the fee. The pack is yours to copy inside your organisation. We keep a copy of the classroom version of the prompt file (the one written on the demo pack) for the next open class. We do not keep a copy of any file you brought that still contains your operational content.

The 30-day clinic is for questions that appear when the method meets a live queue. It is not extra teaching of a different task, and it is not a support desk for the vendor. If the clinic uncovers a gap in the prompt file, we patch the file and send it back. After 30 days the slot lapses; a later question can be booked as a new scoping call.

Class size, language, and the room

Twelve is the maximum for Foundations, forecast, reporting and policy. Ten is the maximum for inbox and team-lead programmes, where each person needs more live review. Below six, an open class is postponed to the next listed 2026 month, with enrolled people offered a move or a refund.

Teaching is in English. Materials are in English. We can read a mix of English and common Singapore office shorthand in sample tickets; we cannot run the class in another language in 2026. The room is at 1 Fusionopolis Place, #08-11, Galaxis, with step-free access from the lobby lifts. Tell us on enrolment if a learner needs a printed copy at a larger type size: we will have it on the table.

Bring a task to the next call

If you already know which of the six programmes fits, say so in the form. If you do not, describe the weekly work in a few sentences: volume, tools, and whether a stripped sample can exist. We will answer with a named programme or with a clear no.

Contact and enrolment