Processing Email Attachments

Processing Email Attachments

Attach a PDF or image to your email and any modern text agent on Pro or Power will read it natively — no special "vision" or "document analysis" agent required.

Attachments are a Pro+ feature

Free accounts get text-only conversations. If you email an agent from a Free account with a PDF or image attached, the AI replies based on your email text alone and the response carries a one-line note explaining the gate. Vision and document tokens cost materially more than plain text — gating them keeps the Free tier sustainable.

Upgrade to Pro to unlock attachments.

What's actually read

Currently read natively (Pro+):

• PDF documents — .pdf
• Images — JPEG, PNG, WebP, GIF (screenshots, diagrams, scanned notes, whiteboards, photos)

Not yet supported:

• Word documents (.docx), Excel (.xlsx), CSV (.csv), plain text (.txt), source-code files and other formats

If you attach an unsupported file, the AI still replies, and the reply includes an explicit "Attachment not read" notice naming each file that was skipped — a paying customer will never silently lose an attachment. To work around this today, paste the relevant text into the email body, or convert the file to PDF before sending.

Native ingestion for Word, Excel, and CSV is on the roadmap — it requires server-side text extraction, which we'll add once the PDF and image flow has settled.

How attachments are processed

When your email arrives, SlyReply:

1. Identifies which attachments are natively supported (PDF + image)
2. Forwards them to the AI agent alongside your email body — base64-encoded, in the same API call
3. Notes any unsupported files in the reply so you know exactly what was and wasn't read

The agent sees both your message and the supported attachments, so you can reference them naturally:

To: lizzy-the-lawyer@slyreply.ai
Subject: Review attached contract

Hi Lizzy, please review the attached vendor agreement
and flag any clauses that are unusual for a standard
SaaS subscription contract. Pay special attention to
the liability cap in Section 7.

Limits

The size cap is the total of all attachments on a single email, not per file.

Inline images in the email body (signature logos, screenshots embedded in the message) don't count against either limit.

For larger documents, send the most relevant pages or sections. Attachments beyond the limit are dropped before the AI call and noted in the reply.

Best practices

• Reference the attachment in your email body — tell the agent what to look for. A focused ask gets a sharper answer than an open-ended "summarize this."
• Name files descriptively — vendor-agreement-acme-2026.pdf beats doc1.pdf.
• One document per email works best for detailed analysis; send multiple only for comparison tasks.
• Big attachments take a little longer — a large PDF is more for the model to read, so the reply takes slightly longer to land. There's nothing extra to pay: billing is flat-rate.