Guide
How to use AI for content creation
Draft from a bounded set of notes, verify each claim and preserve the sources behind the published piece.
By tentus team · Updated 2026-10-26 · 1 min read
Define the reader and the source set
Write the question your piece answers and who needs it. Choose a few saved notes or documents you have permission to use. Record their dates and links; connected accounts are not a guarantee that Lumi can retrieve all their contents.
Build an outline with evidence slots
Ask Lumi: "Using these sources, propose an outline. For each section, identify the supporting source and mark unsupported claims as needing evidence." Review the source behind each proposed point before accepting the outline.
Worked example: a client onboarding article
Fictional example: a 12 October note records repeated setup questions, and a 19 October note proposes a checklist. You can describe the problem and proposed checklist. You cannot claim that the checklist improved retention: neither note contains an outcome measurement. Label the proposal and leave the result unknown.
Review the draft before publishing
Check quotations, numbers, dates and permissions against the originals. Remove unsupported claims, distinguish your advice from observed results, and keep an internal source list. Publishing and client approval are separate actions; generating a draft does not perform either.
Put this into practice
tentus makes it practical. Capture, connect, and retrieve with Lumi working in the background.
More guides
How to use AI for meeting notes
A practical, no-bot workflow for recording a meeting, checking the transcript and generated briefing, and carrying the decisions into your next step.
How to use AI for research
Build a small, checkable research note with source links, competing claims and explicit gaps.
How to use AI for journaling
Review your own dated entries with optional AI assistance, while keeping interpretation separate from what you recorded.