Packback Discussion Bot — AI Posts & Replies

What it handles

Packback, done for you.

Written as a person thinking

Posts are composed against the professor's actual prompt rather than rephrased from a template — varied sentence length, contractions, first-person voice, and the stock phrasing models reach for stripped out.

A voice pass on every post

Each question and reply goes through the writing voice pass before it is submitted: varied cadence and plain wording rather than uniform AI rhythm. Quotations, citations, formulae and code are excluded from that pass, so nothing inside a quotation mark is ever reworded.

Curiosity-score aware

It reads the minimum curiosity score out of the instructor's prompt and writes to clear it with room to spare, targeting roughly fifteen points above the required minimum. Packback assigns the actual score.

Real replies to classmates

It reads the live discussion feed, skips anything you posted this run, and replies to genuine classmate posts — favouring the ones with the fewest existing replies, so your participation is not all piled onto the same popular thread.

On-topic, on-prompt

Questions and replies are generated against the professor's actual prompt and your course context, so they stay substantive and aligned instead of generic filler.

Your voice, if you give it samples

Add a few pieces of your own writing and Silent Student builds a style profile, so posts match how you already write across the course rather than reading like a stranger. Without samples it writes in a plain, natural register instead.

How it works

From open tab to submitted.

1

Point it at the assignment

Silent Student launches Packback through Canvas, confirms the LTI handoff, and reads the discussion requirements — the question count, the reply count, and any minimum curiosity score.
2

It writes and runs the voice pass

It drafts the question and the replies against the prompt, then runs each through the voice pass. Because Packback posts publish directly, this is the point at which the wording is settled — there is no approval step after it.
3

It posts and replies

It submits your question, reads the live feed, and replies to real classmate posts one at a time, retrying or skipping a form error without losing the whole assignment.
4

You read the log

The run finishes on its own and logs exactly what it posted, so you can go and read your own contribution after the fact — and edit it in Packback if you want to.

Where it stops

What it won't do.

Packback behaves differently from the rest of the product, and the difference is worth reading before you queue it.

There is no approval step

Canvas essays, discussions and file uploads are drafted and held. Packback posts are not — they publish directly during the run. If reviewing before publication matters to you on this assignment, do not queue it and walk away.

No AI-detector promise, in either direction

The voice pass is about readability and matching your own writing, not about defeating a checker. No tool can promise a specific AI-detector outcome and we do not claim one. Packback's own scoring is its own.

The curiosity score is a target, not a result

It writes to clear the instructor's minimum with room to spare. Packback assigns the score after the fact, and we cannot promise what it will be.

It depends on the Canvas LTI launch

Packback is reached by following your Canvas assignment link through the LTI handoff. If that handoff fails — commonly a stale Packback session — the run reports the failure rather than posting something half-formed.

Questions

Packback FAQ

No, and this is the one place Packback differs from Canvas. Canvas essays, discussion posts and file uploads are always drafted and held for your approval. Packback posts publish directly as part of the run. If you want to see the wording before it is public, Packback is not the assignment type to queue and walk away from — check the activity log, and edit in Packback afterwards if you want to.

It reads the professor's prompt and your course context and writes against them, then runs the text through a voice pass — varied sentence rhythm, plain wording in place of the stock phrasing models reach for, and your own style if you have added writing samples. Quotations, citations, formulae and code are excluded from that pass. No tool can promise a specific AI-detector outcome, and we do not claim one; you are responsible for following your school's academic-integrity policies.

Both. It reads the live discussion feed, filters out anything it posted this run, and replies to genuine classmate posts. It prefers posts with fewer existing replies so your participation is not all piled onto the same popular thread.

It parses the minimum curiosity score out of the instructor's prompt and aims to clear it, targeting roughly fifteen points above the required minimum. Packback assigns the actual score, so that is what the writing aims at, not a number we can guarantee.

Yes, if you give it something to work from. Add a few of your own writing samples and Silent Student builds a style profile it applies to your posts. Without samples there is nothing to match, and it writes in a plain natural register instead.

More integrations