Writing tasks and process evidence for education
Humanly lets a teacher or task owner configure a writing task, provide instructions and resources, and inspect submitted writing with its recorded process. Task controls apply inside Humanly; they cannot enforce rules on other websites or devices. The record supports a discussion about how the work developed under the task's rules. It does not make an automatic academic-integrity decision.
Configure the task before students write
Use the publisher workflow in the Humanly application to create an assigned task. Define the writing instructions and any resources, then choose the controls that matter for that assignment. Supported control categories include AI access, copy and paste, resource downloads, timing, submission length and attempts. The available setup determines the workspace students enter.
Publish and distribute the task using the offered invitation or public-link flow. Tell students what is recorded and who will review it before asking them to begin. Keep the task's written expectations consistent with its controls: a task that allows AI polishing should not later treat any AI-assisted revision as disallowed.
For example, an instructor can ask for a response to a supplied reading, allow source consultation, and permit polishing while reserving the argument for the student. The written instructions should explain what “polishing” means for that assessment. A configuration is useful only when participants also understand the assessment criteria.
Select an AI rule that matches the learning objective
Humanly exposes None, Read Only, Polish Only and Full AI access modes. The modes govern the in-platform assistant's available context and editing capabilities. Optional custom rules and web-search controls further shape the configured assistant. See AI policy for the precise distinction.
These controls apply to the Humanly workspace. They do not stop a student consulting an external AI service, another device or another person. If an assessment requires a broader rule, communicate that rule separately; do not describe the platform setting as proof that it was followed everywhere.
How can I review an AI-use concern using process evidence?
Start with the task's stated permission and a specific question about the writing. Humanly evidence can describe recorded actions; it cannot decide whether a student violated a policy.
- Read the submitted response and identify the passage or action that needs explanation.
- Inspect the writing setup recorded with the evidence. Distinguish prohibited assistance from permitted polishing or source discussion.
- Check what evidence is available. A shared certificate may hide edit history, and writing completed elsewhere has no earlier Humanly history. Missing or inaccessible history is not proof of AI use or misconduct.
- Use the activity timeline and available replay to examine how the passage changed. Compare final-text composition with process input volume; neither is a percentage of “human effort.”
- Consider the writer's explanation, citations and other relevant drafting context alongside the record. Treat enabled behavior analysis as a prompt for review: a flag is not a finding of misconduct, and an unavailable result is not a clean bill of health.
A pasted passage may be a required quotation or a policy violation depending on the assignment. A period when the workspace is hidden does not identify what site or application the student used. Read the certificate guide before drawing conclusions from either signal.
Plan access and student communication
Process records can contain drafts and AI interactions that are more revealing than a final submission. Decide who needs to see that material and explain the review process, including how students can provide context. Institutional requirements for consent, retention and appeals need an explicit local policy; the reference pages do not establish those terms or promise an institution-specific compliance certification.
The privacy policy describes the service's handling of data and AI features. The research guide distinguishes published usability and detector studies from evidence of classroom learning outcomes. Those studies do not establish that using Humanly raises grades or reliably resolves every authorship dispute.
Reviewed: 2026-09-24.
Applies to: Humanly writing workflows reviewed on 2026-09-24. Available controls and evidence depend on the writing setup. Research findings refer to arXiv:2607.21758v1.
Canonical: https://writehumanly.net/reference/education