MktPraxis home
Two office hours a week

A teaching assistant for every student

You have ninety students and two office hours. Four people turn up, and they are rarely the four who needed to. The scaffolding here works the other eighty-six, at the moment each of them actually gets stuck rather than the moment your calendar allows.

The student who needs you most is the one least likely to email you. They stall at 11pm on a Tuesday, conclude they are simply bad at statistics, and quietly stop opening the tool.

Something opens beside the tool and refuses to move on

Professor Mode spotlights the one control that matters next and will not advance until the student has actually used it. Watching is not an option, which is the whole point.

PROFESSOR MODE

It also stops to check understanding partway through, so a student cannot click their way to a finished analysis they could not reproduce on Monday.

Three kinds of help you are not personally delivering

The walkthrough, for the student who is lost

Step by step, inside the tool, with the relevant control highlighted. It handles the eighty percent of office-hours traffic that is really "I do not know which button does the thing."

Praxis Pal, for the student who is stuck

It can see the file they uploaded and the output on their screen, so it answers about their variables. It explains and prompts; it will not write their memo, and it never hands over a finished interpretation to paste in.

Feedback that teaches, for the one who got it wrong

Quiz feedback names the misconception behind the wrong answer rather than announcing a letter. A missed item becomes a second chance to learn the idea instead of a small punishment.

What that adds up to

55 guided walkthroughs, hand-written for the specific tool they belong to
63 tools with the assistant already open in the corner of the page
24/7 availability, which two office hours a week has never been able to offer

⚠️ The fair objection

"This sounds like it teaches for me, and badly."

It teaches the mechanical layer — which control, which assumption, why that output is not an error. That layer is what consumes office hours and what students are most ashamed to ask about, and it is exactly the part that scales. What it deliberately does not touch is the judgement: what the finding means for the brand, whether the sample earns the claim, what you would actually recommend. That still needs you, and clearing the mechanics off your desk is what buys the time for it.

And yes, the quizzes auto-grade and the gradebook reconciles overrides with a visible trail. That is worth having, but every platform does it now, so it is not the reason to be here.