LifterLMS Assignments are graded tasks attached to lessons. The friction is rarely the writing, it's the interpretation. Students re-read the prompt three times, lose track of word counts, miss the rubric weight on each criterion, and submit work that scores low because they answered a question that wasn't actually asked. A generic chatbot has no idea what the assignment says and ends up giving boilerplate study tips.

SleekAI maps the llms_assignment post fields, including the task description, the rubric meta, the point value, and the student's draft history from wp_lifterlms_user_postmeta, into the bot's system message. The tutor can then ask the right follow-up questions, point out missing rubric criteria, and suggest a stronger thesis statement based on the actual prompt. Conditional display means the assignment tutor only appears on assignment lessons, not on regular reading lessons.

Generic bots also flatten subject matter. A creative writing assignment needs different feedback than a coding exercise or a case study. SleekAI lets you run different bots per course, each with its own system message tuned to the subject, while sharing one OpenAI or Anthropic key across the site. Conversation logs tag each message with the assignment ID so instructors can see where students struggle and update the prompt copy directly.

The hardest part of any assignment is not the writing, it is the interpretation. A student who misreads the prompt produces a clean essay that misses the point and scores low, and that feedback loop is slow and discouraging. Office hours are once a week, instructor email is once a day, and the assignment is due tomorrow.

A generic chatbot fills that gap with vague study tips because it cannot see the prompt or the rubric. SleekAI changes the dynamic by reading the actual assignment post and the student's current draft into the system message at request time. The tutor can then point to weighted rubric sections, suggest a thesis statement that matches the prompt, and warn when the draft is short of the word count range.

Conversation logs tagged by assignment ID give instructors a view into where students struggle, which often turns into clearer prompt copy and better learning outcomes the next cohort around. Everything runs on the instructor's own provider key, so the model choice, the cost, and the data control all stay in their hands.