With the Document Reader you have GENI read a web page or a block of text and answer from it: a summary, the key points, or the answer to one specific question.
Available from Starter. No connection needed.
What can GENI do with it?
- Analyse a document or web page from a URL or from text you paste into the chat, answering the question you attach to it
One tool, but your question decides what you get back: summarise, pull out the key points, compare two passages or look up one thing.
Example: what you ask, what you get
Summarise these terms and tell me what stands out
[text pasted]
Summary , Supplier X terms and conditions
- Payment term 14 days, then 2 percent interest per month
- Notice period three months, only at the end of a quarter
- Liability capped at the invoiced amount of the last 6 months
What stands out
- Article 9 allows a unilateral price increase with 30 days notice
- There is no agreement on delivery time anywhere in the document
And from a URL:
https://example.com/research-2026 , what does it say about small business?
The report mentions small business in two passages. The gist: 34 percent use AI weekly, against 71 percent at companies above 250 employees. The reason given is not price but the time it takes to set up.
Source: example.com/research-2026
Requirements
- Plan: Starter and up
- Connection: none
Activating
- Go to Dashboard -> Skills and activate Document Reader
- Paste a URL or the text into the chat and ask your question
A PDF is genuinely read
Give it a link to a PDF and the file itself goes to the model. That is something other than squeezing flat text out of it: tables stay tables, columns do not run into each other, and a scanned document is simply read. Exactly the difference between reading an invoice and guessing at it.
You do not have to switch anything on. If GENI sees the link is a PDF, it takes that route automatically.
What it costs
| Action | Credits |
|---|---|
| Analyse a web page or pasted text | 5 |
| Read a PDF | 10 per page, 25 minimum |
A single-page PDF therefore costs 25 credits, a twenty-page report 200. You pay for what the model actually has to read. GENI mentions the price before opening a PDF. The conversation itself comes on top; see The credit system.
Or send the file along. You can also attach the same PDF in the chat (paperclip, up to 5 MB). That route does not run through this skill but through the conversation itself, so you pay the normal conversation cost instead of a per-page price. A link is handy when the file lives online; uploading is handy when you have it on your own computer.
Limits
- PDF up to 25 pages and 10 MB. Anything larger is refused with the reason, rather than read halfway. Split the document or say which part you mean.
- Want to send a file instead of a link? Use the attachment button in the chat , that runs through the chat itself, not through this skill.
- Chat attachments: images, PDF, txt, markdown and csv, up to 5 MB per file and 3 files per message.
- A page behind a login or paywall cannot be fetched. Paste the text instead.
- Internal or private addresses are blocked. Publicly reachable URLs only, which is a deliberate safety measure.
- Very long documents are read partially. Ask a specific question or supply the relevant section.
- It does not store the document. To search it again later, use the Support Knowledge Base.
Troubleshooting
"Failed to fetch URL". The page blocks automated visits or asks for a login. Paste the text directly into the chat.
"No usable content found". The page builds its content with JavaScript, or the PDF sits behind a viewer instead of the link pointing at the file itself. Copy the text across, or give the direct link to the .pdf.
"This PDF has roughly X pages". Above 25 pages it will not read it in one go. Split the document, or ask for the part you need.
The answer is too shallow. Ask a sharper question. "What does it say about the notice period" yields more than "summarise".
I want to find the document again later. This skill cannot do that. Put it in the Support Knowledge Base or keep the summary as a note in your Workspace.
Frequently asked questions
Can I have a PDF read?
Yes. Give a link to the .pdf and the file itself goes to the model, with tables
and scans intact. That costs 10 credits per page, up to 25 pages. If you would rather send
a file than a link, use the attachment button in the chat (up to 5 MB); that runs
through the chat, not this skill.
Why is a PDF so much more expensive than a web page? Because there is far more to process. A page of text is a few thousand tokens, a 25-page PDF well over thirty thousand, and with a PDF the layout comes along too so that tables stay correct. You pay for what the model actually reads.
What is the difference with the Support Knowledge Base? This skill reads one thing, now. The Knowledge Base indexes your documentation so you can still search it months later.
Does it work with a Google Docs link? With a publicly shareable link, yes. If the document is restricted, connect Google Drive.
Can it compare two documents? Yes, paste both and ask for the difference.
Back to Skills marketplace See also: Support Knowledge Base · Google Drive · Workspace On the site: all skills
dGENIX Docs, Document Reader, updated August 2026