A photographer opens a feed to research lighting and leaves with twenty product reviews. The topic was close; the intent was wrong. An interest brief describes the work you want to learn, the examples you need, and the material you want less of. Write it before adjusting a feed or asking an assistant to organize inspiration.
Meta's October 13 Facebook announcement introduces natural-language interests for Feed and Reels through Your algorithm and post feedback. It says the tools are rolling out; check your account's three-dot menu rather than assuming access. This October 14 guide applies that news to an editorial research habit. It does not promise wider distribution for a creator's posts.
A filled interest brief for a real decision
This campaign example is fictional. Its constraints illustrate the brief, not measured results or a claim about a particular feed.
| Field | Filled example |
|---|---|
| Current project | Plan a short tutorial on photographing handmade ceramics at home. |
| More useful topics | Window-light setups, low-cost reflectors, photographing glaze, and explaining lighting to beginners. |
| Intended audience | Makers using a phone camera in a small room. |
| Preferred evidence | A visible setup, before-and-after images, and an explanation of what changed. |
| Less useful material | Camera shopping, studio tours with no technique, and unsupported claims about going viral. |
| Output from research | Three original tutorial angles with a demonstration we can shoot ourselves. |
| Review boundary | Keep source links; do not copy a creator's script or infer permission to reuse their images. |
The useful sentence is more specific than “show me photography.” It connects a subject to a job. A beautiful studio can be interesting without answering the beginner's small-room question.
Separate what you watch from what you publish
Your personal recommendations are a place to find examples. They are not a representative survey of everyone who might watch your work. Record the audience decision separately: who needs the tutorial, what equipment they have, and which problem should be solvable afterward.
Ask an assistant to turn the brief into questions before asking for ideas. Which surface creates glare? What can the maker change without buying equipment? Which two pictures would show the effect clearly? Those questions give research a purpose and prevent a list of fashionable topics from becoming the campaign plan.
Describe the unwanted match precisely
“Not photography reviews” may discard useful comparisons as well as shopping content. Use the reason: “I need lighting demonstrations that show the setup; camera buying advice does not help this project.” Keep the wording about the task instead of judging the person who made the post.
Meta's earlier explanation of AI ranking describes multiple predictions and signals working together. No single prediction is a perfect measure of value. Treat that as background for careful observation, not a formula for controlling recommendations.
When a recommendation misses the brief, note whether the subject, skill level, format, or intent missed. Change one part of your wording, then inspect what appears. A few examples can show whether the adjustment is useful for your session; they cannot establish an algorithm-wide effect or predict how your own content will rank.
Keep an evidence notebook beside the feed
For each useful example, save its link, the date you viewed it, the technique it demonstrates, and the question it leaves open. Give each saved example a purpose.
A good note might read: “The creator moves a white card closer to reduce a shadow. Test a similar setup with our own ceramic cup; check whether glaze reflections obscure the rim.” This is an observation and a proposed experiment. It does not claim the original creator's method always works.
Write your own shot sequence from the question. For the fictional tutorial: show the untreated scene, move the reflector, keep the camera position fixed, and explain the visible difference. Review the result before writing a broad claim. If the change is barely visible, revise the demonstration instead of exaggerating the caption.
Give AI a bounded transformation job
Use the notebook to ask for a planning artifact: “Group these approved notes into three tutorial angles. For each, give the audience question, a shot we can capture, and the evidence we still need. Keep each source attached to its observation.”
Review whether the output preserved the audience and the uncertainty. Return an angle that requires equipment you do not have or describes an untested technique as proven. If several tools produce plans, use an agent evaluation task pack to compare their handling of the same materials rather than their confidence.
The workflow can be adapted to Growit o1. Check o1 guides for the platform, account, and release availability of the tools you intend to use. Another company's feed controls do not establish an o1 integration. A planned agent voice call or preview is not a verified live connection; the physical o1 device remains in development, as described in o1 specifications.
Turn one useful example into your next artifact
Fill the brief for your current project, inspect the controls actually available in your account, and collect three examples that answer a defined question. Produce one shot list or draft outline from those notes. Keep the source observations, your proposed changes, and the result you tested separate.
Growit's AI agents guide helps frame the handoff. Start with a concrete output a person can inspect. Use recommendations to plan your own experiment and explain the result.