Start without connecting apps
Do you ever have a document or receipt whose small print is difficult to read in full? Gemini can read text directly from a photograph as part of a normal conversation. The “Double-check response” feature below the response compares it with sources from Google Search.
Your goal for this session
Take a photo of a real document or object and ask Gemini to read and organise its contents. You do not need to give it access to your private information, and you will also learn when using an image helps.
How to make your first attempt
- 1
Open Gemini
Open Gemini in your browser or the app and sign in if the service asks you to.
- 2
Choose an image containing text
Take a new photo or find one in your camera roll with clearly legible text: for example, operating instructions, assembly instructions or text on packaging.
- 3
Attach the image and send your request
Press the attachment icon in Gemini's message field and choose your image. Then copy the request below into the same message and send it.
- 4
Check the response against the image
Read Gemini's response and compare it with the image. If anything is wrong or unclear, point it out and ask for a correction.
Opens in a new tab. Then return to this guide and copy a request to go with the image.
Open Gemini and try it (opens in a new tab)Read the image and organise its contents
Here is a photo of a document that I find difficult to read in full. Go through the text in the image and organise its contents into clear, numbered points. If any word or number is unclear to you, do not guess: tell me where it appears in the image and what you are unsure about.
Compatible with: Gemini
Refine the response using the same image
Ask for a closer reading
A common request:
You get a general summary of the image, but one point is still unresolved.
Try asking this instead:
Go back to the third point in the image and quote its exact wording as it appears, instead of summarising it.
The refinement directs Gemini back to one part of the image, so a reading error becomes immediately visible instead of getting lost in the whole response.
Two more ways to use images
Try these two image tasks. Before uploading a receipt, inspect it and remove personal, payment and other sensitive details, including details in barcodes or QR codes. You can use a fictional or public sample receipt with no sensitive details instead. Attach the image to a Gemini message, then send the request. After the plant-care response, use the “Double-check response” control yourself if it is available. Inspect its findings and open the linked sources to check whether they support the care instructions; the check does not guarantee accuracy.
Turn a shopping receipt into a table
Here is a photo of a shopping receipt. List each purchased item and its price on its own row in a table, then calculate the total of the rows. Compare the sum with the receipt's final total and tell me whether they match. If a row is partly hidden, mark it “unclear” and do not invent a price.
Compatible with: Gemini
Identify an unknown plant — and check against current information
Here is a photo of a houseplant whose species is unknown. Identify the plant to the best of your ability and give the three most important care instructions: light requirements, watering frequency and one other consideration. If you are unsure of the species, say so directly and suggest the two most likely options.
Compatible with: Gemini
Once you know the basics: connected apps
Once you are comfortable using images, you can explore Gemini's connected apps. They let Gemini use services such as Gmail, Calendar or Drive directly as part of its response. Availability depends on your country, device, account and the version of Gemini you use.
Connect one app at a time
When you want to try a connected app, first check in the settings what information it uses, and grant permission only for a task you actually need. You can disconnect it at any time.
When might Gemini suit you?
Gemini may be useful if you often handle documents, receipts or other real objects that you want a quick summary of. The same image recognition also works with handwritten notes and screenshots, as long as the text is clearly visible in the image.
Show an image. Ask Gemini to read it.
Gemini can read images, not just text. Choose a situation, attach an image and follow up with one refinement.
Your first try · about 3 min
Where shall we start?
How the request is built
- Goal
- Explain the care label's symbols in plain English
- Format
- Three points: washing, drying, ironing
- Check
- Point out an unclear symbol and do not guess the temperature
- You provide
- A photo of a clothing care label, attached to the message
Here is a photo of a clothing care label. Explain its symbols in plain English under three points: washing temperature, whether tumble drying is allowed and whether ironing is allowed. If any symbol is not clearly visible in the image, say so directly and do not guess the temperature.
Ready to try it in the actual Gemini?
Copy the request first. Gemini opens in a new tab, and this exercise will be waiting for you here.
Open GeminiOne useful rule: Always check important numbers, names and words against the original image before acting on them.
Publication details and sources
- Published:
- August 12, 2026
- Sources checked:
- September 4, 2026