FliesReplies Now Reads the Image, Not Just the Caption

Open LinkedIn right now and scroll your feed. Count the posts. About 7 out of every 10 will have an image, a screenshot, a chart, a carousel, or a PDF document attached. The caption above each image is usually short — sometimes just a tease ("worth a read") or a one-liner. The actual content lives inside the visual.

Every AI reply tool you have used until today reads only the caption. That is why so many AI-generated replies feel generic — the AI is replying to "worth a read" without ever seeing what is worth reading. As of today, FliesReplies reads the image too.

What This Actually Means

When you click the FliesReplies button on a LinkedIn post that contains an image, the extension extracts the image, sends it to the AI alongside the caption, and the AI reads both before generating a reply. Replies now reference what is actually visible in the image — the numbers on a chart, the text overlaid on a meme, the headline of a screenshotted tweet, the specific products shown in a launch photo.

A Real Example

A post on our feed yesterday had the caption "Tim Apple is back…incredible" — six words. The attached image was a screenshot of a Truth Social post listing a dozen CEOs flying to China on Air Force One: Jensen Huang, Elon, Tim Apple, Larry Fink, Stephen Schwarzmann, and several more.

A text-only AI reply tool sees only "Tim Apple is back…incredible" and produces something generic about the Tim Apple meme. FliesReplies sees the image and produces: "bruuuuh 'Tim Apple' will never not be funny to me — the man is literally on Air Force One name-dropping CEOs like a festival lineup." That reply names "Air Force One" and "name-dropping CEOs," details that only exist inside the image. The original poster knows this reply read the actual post, not the caption alone.

Which Post Formats Are Supported

• Single image posts — fully supported, replies reference image content
• Image carousels (3–10 swipeable images) — the first slide is read, which is usually the hook
• Document and PDF posts — LinkedIn renders each PDF page as an image, so the cover slide is read
• Screenshot posts — tweets, charts, headlines pasted as images all work
• Quote reposts — both the outer commentary and the quoted original post are read together
• Video posts — replies fall back to caption-only generation (single frames from videos are misleading)
• Text-only posts — work the same as they did before

Why This Matters for LinkedIn Growth

On LinkedIn, the most engaged posts are usually visual. Carousel posts, charts, and PDF "tipfluencer" decks dominate the feed because they get more dwell time. If your reply tool cannot see those visuals, your replies on the highest-engagement posts will read like you skimmed the title and moved on. Image-aware replies make engagement feel intentional — like you read the chart, understood the carousel, and had a take that maps to the actual content.

No New Steps for You

You do not need to enable anything. There is no toggle, no upgraded plan, no new button. The next time you generate a reply on an image-heavy post, the AI reads the image automatically. Your replies just get sharper.