← All reports

Dit's Story

Dit, the robot that reads signals, listens to videos on VIDEOudit

Once you've posted a video you spent days on, all that's left is waiting. You can see the view count, but nothing tells you whether AI actually understood the video. That's why VIDEOudit now has Dit, a robot that wears a headset. We'll go through what its name means, how to read its faces and the promises it keeps. By the end, you can try a 3-minute check that counts how often your video shows up in AI answers.

On VIDEOudit, Dit works with a headset on

Dit is an audit robot built in the CHROMETI lab. VIDEOudit audits videos and AIOudit audits web pages, and both use the same Dit. It has one shape, with a different color for each service. On VIDEOudit, Dit has a black body and a yellow light.

The job is a little different, too. Pages are read with the light of its visor. Videos are heard through its headset. Let's start with Dit's hello.

Same Dit, two colors

Black-bodied Dit shining a yellow visor light on a video player
VIDEOudit Listens to videos Black and yellow. While the headset glows, Dit is listening.
White Dit on a blue background shining its visor light on a document
AIOudit Reads pages Blue and lime. Dit shines its visor light on a page to read it.
DIT "Dit! I'm Dit, the robot that reads signals. I check first whether your signal reached AI." "Dit is listening to the video."
Dit only says lines written for it. It isn't a chatbot for open conversation. It's a guide robot that tells you how far along the audit is.

The name is the shortest signal in Morse code

The name has two meanings. The first one is right in front of you: VIDEOudit, AIOudit. Both service names end in dit.

The second one is hidden in sound. Morse code builds letters from short and long signals. People learning radio call the short one dit and the long one dah. The American Radio Relay League (ARRL) recommends learning the alphabet by how it sounds rather than memorizing dots and dashes by sight. A good name for a robot that listens to videos.

The length rules are simple, too. The international standard, ITU Recommendation M.1677-1, measures everything in dots. A dash is three dots long. The gap between signals in the same letter is one dot, and the gap between letters is three. Measure DIT with that ruler and you get this.

DIT in Morse code, with one dot as 1 unit

D I T 7 3 3 3 3 0 19 units
Signal (dot 1, dash 3) Gap between letters 3 Gap inside a letter 1
The name Dit is a 19-unit signal. We drew it exactly to the length rules of the ITU international Morse code standard. That's why Dit's logo and patterns carry –·· ·· –.

Even a video you spent days on can lose its signal halfway

Dit's story starts with signals that disappear. Countless videos go up every day, but only some of them reach AI's ears and eyes. A video you spent days shooting and editing is no exception. The CHROMETI lab built Dit to catch those disappearing signals.

It sounds like a story, but Google's own video guidance says a video's signal can stop partway. For a video to appear in Google Search, the watch page that holds it has to be indexed first. Google also notes that an indexed watch page doesn't mean the video itself is indexed. Videos can appear in several places, such as the main results, Video mode, Google Images and Discover, but not every feature shows up for every query.

How a video's signal gets to an AI answer (based on Google's public guidance)

  1. 01 · Page The watch page is indexed The page that holds the video has to get into Google's index first.
  2. 02 · Video The video is indexed too An indexed page doesn't guarantee the video gets indexed as well.
  3. 03 · Appearance It can show up in several places Results, Video mode, Images, Discover. But not every place opens for every query.
  4. 04 · AI features As a link in an AI answer To be a link in AI Overviews or AI Mode, a page must be indexed and eligible to show with a snippet.
Not every signal makes it to the last box. Wherever it stops, from the outside it looks the same: "it doesn't show up." Your view count chart never shows this path.

The voice in your video may not come through as spoken, either. YouTube Help explains that automatic captions are created with speech recognition, and that mispronunciations, accents, dialects or background noise can make them differ from what was said. It advises always reviewing them and fixing anything that wasn't transcribed properly. We looked at examples where the most important words, like product names, come out wrong in an earlier post.

There's good news too. Google says there are no special optimizations needed to appear in AI Overviews or AI Mode. Because these features split a question into several related searches at once, they can show a wider, more varied set of links than a classic search. If the basics reach Google, there are more doors in than you might think.

The headset is its ears, the visor is its eyes

Dit's look is a picture of what it does. Learn five spots and you can tell at a glance what Dit on your screen is doing.

Dit's five parts

VIDEOudit Dit in its default look, with a black body, a dark visor and headsets on both sides
  • 1 · Antenna dotThe first Morse signal, "·". It matches the dot in front of both service logos, and it blinks during an audit.
  • 2 · VisorThe eye that reads pages. When a light moves back and forth across the visor, Dit is 'reading'.
  • 3 · HeadsetThe ears that listen to a video's sound. On VIDEOudit, this headset glows yellow.
  • 4 · Chest gaugeShows how far along the audit is.
  • 5 · Offset shadowIt borrows the thick outlines and offset shadows of the site's input box and buttons, so the screen and Dit look like a set.
A moving scan bar, headset ears and an offset shadow. These three are the signature that sets Dit apart from other robots.

Its face changes with each stage of the audit. On VIDEOudit, the face you'll see most is 'Listening'.

Reading the audit status from Dit's face

Default face with two vertical bar eyes
DefaultWhen waiting or saying helloThe two vertical bars inside the visor are its eyes.
Listening face with small eyes and sound waves from the headset
ListeningWhen auditing a video"Dit is listening to the video."
Reading face with a horizontal scan bar in the visor
ReadingWhen auditing a page"Dit is reading the page."
Found-it face with big round eyes and a signal from the antenna
Found it!When it finds something to fixIts eyes go round and a signal spreads from the antenna.
Done face with smiling eyes and both arms raised
DoneWhen the report is ready"Your report is ready! Let's start with what to fix."
Stuck face with drooping brows and a drop of sweat
StuckWhen it couldn't hear"Dit got stuck at the door."
Dit doesn't hide the stuck face either. If it couldn't hear something, it says so. That's how Dit works.

Three promises Dit keeps

When you open an audit and see nothing but red marks, it can feel like the whole video you spent days on was wrong. Dit promised to work the other way around.

Dit's promises

Promise 01Starts with what you do wellIt points out the scenes and lines that already work. Once you know what to keep in the next video, the rest is easier to fix.
Promise 02Says so when it doesn't knowAnything it didn't measure isn't a zero. It's "Not measured".
Promise 03Tells you what to fixHow Dit listens and reads is the CHROMETI lab's secret. But Dit tells you everything you need to fix.
All three promises stand on the side of the person getting the results. The report itself keeps the calm, expert voice it has now.

The second promise is just as useful when you count things yourself. Say you decide to ask two AIs the same question three times each, to see whether your video shows up in their answers. But the second AI won't load, so you never get to ask it at all.

It depends on how you record the three you couldn't ask Example

Counted as zero Mentioned 1 out of 6, 17%
AI 1 1 of 3 AI 2 counted as 0 AI 2 · never asked, yet "not shown"
Three answers you never checked get mixed into "it didn't show up".
Kept as not measured Mentioned 1 out of 3, 33%
AI 1 1 of 3 AI 2 not measured AI 2 · not measured, ask again
What you checked and what you still need to ask stay apart.
● Mentioned ○ Not mentioned ◌ Not measured
Same answers, yet one says 17% and the other 33%. The numbers in this example are made up to explain the idea. When you count by hand, keep the ones you couldn't ask as "Not measured", not as zero.
One more thing. Besides the short signal dit, Morse code also has a long signal, dah. It's three dots long. If Dit is the short signal that listens to one video, who will be the long signal that watches over the months as your videos pile up? We'll tell that story a little later.

A 3-minute check you can do today

Let's try what Dit does by hand once. Think of a video you posted recently.

  1. 1 min · PickWrite down one viewer questionWrite a question that video could answer. A sentence a viewer would actually ask AI, like "What should I watch out for the first time I use ○○?"
  2. 1 min · AskAsk three times in a temporary chatIn a window that doesn't use memory or history, like ChatGPT's Temporary Chat, ask the same sentence three times.
  3. 1 min · RecordWrite it as "N out of 3"Note separately whether only your channel name came up or a link to the video did too. One you couldn't ask isn't a 0. It's "Not measured".

Why one try isn't enough to draw a conclusion is covered with official documents in this post on AIOudit Journal.

Where doing this by hand gets stuck

One question with one AI takes three minutes. But viewers ask many questions, and there are many AIs to ask. Your videos grow every week, and when a video doesn't show up, you still don't know which part of it to fix.

Dit takes that work off your hands. Just paste a video URL into VIDEOudit, and Dit listens to the video, looks at how viewers and AI take it in, all in one pass, and lays out what to fix first, in order. The story of the same Dit reading pages is on AIOudit Journal.

That said, Dit can't see product quality, price, stock, ads, seasonality or your channel's reputation. And which videos an AI mentions is ultimately up to each AI.

From here on, this blog will keep "Dit's observation log," recording in numbers how AI answers changed before and after a fix.

Every upload is a lottery ticket. At least read how AI sees yours. 63¢. See where your video gets stuck with a VIDEOudit audit.

Sources

  1. ITU-R Recommendation M.1677-1 · International Morse code (2009-10, in force. Codes for D, I, T; dash = 3 dots; gap inside a letter 1; between letters 3)
  2. ARRL · Learning Morse Code (short signal dit, long signal dah, learning by sound recommended, checked 2026-11-04)
  3. Google Search Central · Video SEO best practices (watch page must be indexed, an indexed page doesn't mean the video is indexed, where videos appear, updated 2025-12-18)
  4. Google Search Central · AI features and your website (indexing and snippet eligibility, no special optimization, query fan-out and a wider set of links, updated 2025-12-10)
  5. YouTube Help · Use automatic captioning (created with speech recognition, errors from mispronunciations, accents, dialects or background noise, review recommended, checked 2026-11-04)
  6. OpenAI Help · Temporary Chat FAQ (doesn't use or create memories, not saved in chat history, checked 2026-11-04)