Join the beta
A free iPhone app, now in public beta. Example data on this page, except the one real scan marked as such.

See your feed clearly.

Scan a feed you use. Get a plain report: what appeared, which accounts it came from, and which posts carried an ad label.

Join the beta on iPhoneWatch the example film

A scan reads what appeared on screen. It cannot say why any post was there.

One short recording of a feed you already scroll. Review after, not during.

Free. iPhone. Works with Instagram, TikTok, YouTube, X, Facebook and Reddit.

Example filmFour fictional posts, twenty seconds Example
Four posts1 carried an ad label
Fictional example. Review happens after recording.
Read the film in words

An ad label is visible. A political-content classification is an inference. A suggested label does not explain why the post was selected. One readable account appears twice. Four fictional posts become a small record, not a representative feed.

The report

One scan. One plain-language report.

Four dated scans, kept separate. Turn back to any of them. A small scan stays small, and unknown labels stay unknown.

45Posts recorded
42Accounts read
6Printed ad labels
4Scans on record

Counts from the September 20 sheet. Each sheet holds only its own posts.

Turn over the whole example in the explorer
Example recordInstagram. Four fictional scans, Aug 30 to Sep 20 Example
Separate observations

Sep 20, held on record.

45 posts in this scan. 6 carried a printed ad label. 125 posts across 4 scans.

More context does not by itself establish a stable rate or a certified change.

Ad labelUnknownOne mark, one post
One real scan

One real scan, marked as such.

Everything else on this page is a fictional example. This is a single real scan of the founder's own X feed, read by the app on 6 October 2026.

One scan of one feed. It shows what this scan read, not what X shows people in general.

X feed, 6 October 2026Real scan, n=1
36Posts read
5Carried a printed ad label
A single real scan of the founder's own feed, n=1. One mark, one post. Account names are not shown.
Try it

Tap four posts. Watch the record form.

A fictional feed. Each tap adds what a scan could read from that post, and stops at what it could not.

The record forming0 of 4 posts

Nothing yet. Tap Read on any post in the phone.

Illustrative. A scan reads what appeared; it cannot see a platform's model or say why a post was shown.

For youFictional
Record so farnothing yet
Four fictional posts. Tap Read to add one to the record.
@harbor.coffeeSponsored
Fictional example post image: a glass of iced coffee beside a paper bag

A slower start. A fictional sponsored coffee post.

@city.notesFollowing
The local edition

City council votes on a new transit plan.

On the local agenda this week.

@wildpath.journalSuggested for you
Fictional example post image: a still lake at dawn

A long way from the morning rush.

@harbor.coffeeFollowing
Fictional example post image: the same coffee shop, before opening

Behind the counter, before the doors open.

What a scan can read

What a scan can read. What it cannot.

A recording holds what was on the screen. The report stops where the screen stops.

Can read
  • a printed ad label
  • an account name
  • a post's position in the feed
  • the date recorded
  • how many posts appeared
  • an account that returns
  • a suggested label
  • a following label
  • a headline on screen
Cannot read
  • why a post appeared
  • your beliefs
  • your mood
  • the platform's model
  • what you scrolled past
  • which tap led where
  • the whole feed
  • the next feed
  • who else saw it

We show what appeared in a sample and keep observed labels apart from inferences. Political content describes a post, not your beliefs. Tone describes a reading of the content, not your mood. A suggested post does not tell us which action led to the recommendation.

The example on this page is fictional, not a live scan or a demonstration of a running classifier. The one real scan is marked as such. This page does not record your screen or analyze your feed.

Three steps. No jargon.

Click through what happens from a recording to a report. It advances on its own too.

Step 1. Start a scan
InstagramReady to record
TikTokReady to record
YouTubeReady to record
XReady to record
FacebookReady to record
RedditReady to record

Your data stays yours.

A scan handles sensitive material. Here is how it is treated.

Read the privacy policy

Read, then deleted

Frames of your screen are analyzed to build the report, then deleted from your device.

Delete anytime

Deleting your account removes your scan results and profile.

Never sold

No ad tech and no analytics on this site, and your data is not for sale.

Public methodology

Each metric is documented, and the analysis prompts are published in full.

What people do with it

A read on a feed you can act on.

Check your news diet

See which accounts fill your feed, not the ones you think do.

Audit the ads

Which posts carried an ad label, and how many, in each scan.

Re-tune on purpose

Unfollow, mute and follow deliberately, then scan again and keep both records.

Teach and research

Example records for classrooms studying recommender systems.

You shouldn't have to take our word for it.

Each number the app reports is a count with its sample size, documented and open to check.

Public methodologyEach metric documented, the prompts published verbatim.
Honest about limitsWe show what you saw, and say what we cannot read.
No analytics on this siteThese pages set no cookies and request nothing from third parties.
Read the methodology
Worked exampleExample

How a count is made

Printed ad labels, from one fictional scan.

6 of 45posts carried a printed ad label.
Ad label, 6No label, 38Unknown, 1
6 with a printed label, 38 without, 1 unreadable. 45 recorded.
See this count in the explorer

Not on iPhone yet?

One email when AlgorithmLens is on the App Store or Android. Nothing else.